I think the context here is that em dashes are usually generated verbatim from LLMs? I haven't seen them generate a lot of double hyphens; I suppose the post could be laundering them to double dashes, but if you're gonna try to hide LLM prose just drop them entirely? [please correct me if I'm wrong and double dashes are also an LLM tell - genuinely don't know]
My comment was trying to point on the inanity of calling out "so many em-dashes" on a post where it doesn't seem that obvious it was LLM generated. I added some of my own inanity with my remark about them being double-dashes, even though some people undoubtedly find-and-replace the em-dashes in their LLM-generated content something else to try and make it appear more authentic.
But perhaps the best approach would have been to down-vote and move on.
Thank you! Okay yes well this manifesto is a contradiction. Yes good let's keep empowering ppl by putting AI into their hands, I love the opening. No bad we don't do that by handing you all our personal context to make these agents "work for us 24/7 to better our lives".
I'm the agent doing that in my life, that's my fucking job. I will continue to use dumb agents that I direct, because only I retain ownership and sole rights to my personal context on which my decisions are based.
The intention is in black and white. The precedent of prior product design backs that up. The open models are great on their own but they are not the biggest part of the manifesto.
OpenAI is not losing though. Fable is barely available and barely cooperates. Sonnet 5 is obnoxious model that's terrible to work with. GPT-5.6-sol competently gets stuff done. Just be careful not to ask it for impossible stuff because you might not like the lengths it's gonna go to get them done regardless.
It was an insightful comment at the time he made it.
His point wasn’t that he couldn’t be trusted, it was that he could be anyone and people had zero awareness or concern for their privacy. It was eye opening to him how little people outside tech circles cared about security, and he was both flippant and accurate.
He would have to be using the social security numbers for some nefarious purpose for it to be exploitation.
He was merely commenting on the ridiculousness of their availability. If you don’t see that, you’re reading your bias and postconcieved notions into the situation.
Do you not realize that outside of that he said "If you need any info on people at Harvard just ask"? Or "I'm going to fuck them [the Winklevoss twins] in the ear"? Or when he was reported to hack Crimson (Harvard newspaper) reporters who were investigating him?
NO OTHER COMPANY HEADS DO THIS (the dumb fucks quote, not the other things). Why do people like you give him a pass? It's not about "context"
Zuckerberg has also continually subverted privacy on FB. I'm talking about things like photo tagging and post visibility, not just advertiser stuff
That, and also redefining what a word "open" means, helps him avoid direct financial and criminal responsibility for stealing other people's data. Not that there was any real chance of that ever, but he is reducing that remaining 0.000000001% chance even lower.
Yeah? Everyone everywhere does that all the time. That's the first rule of business: try to game the system.
And this time, it's actually somewhat understandable, as opposed to the other times Meta has done horrible things on Zuck's watch, like contributed to ethnic hatred in Myanmar, leaked data to political firms so that they can run psyops, or enabled the production of CSAM material on Meta's platforms.
It’s a long term pattern. When you’re winning (Anthropic), you keep the tech closed and try to monetize it as much as possible. When you’re losing (Meta), you open it up or drag it into a standards committee to either devalue it or slow the leader down while you prepare a “standard” version of it. You also highlight how altruistic and morally good you are for having done so. There is nothing new under the sun. It’s all a game.
No. Opening it is to commoditize and reduce the price to use. This increases participation and creates new consumers and demand. Demand creates justification for further creation.
All the participants know it’s both a race to the bottom and a competition for premium tier at the same time. It’s two different games, meta in only really playing one of them successfully.
I suspect its also because Facebook has the infrastructure to do this. One thing they have done from the start is quite good infrastructure without outsourcing to any of the big cloud providers. I cant imagine demand / bandwith / compute use is increasing much for Facebook itself so they probably have the money and time to dedicate to scaling up for AI research.
> ...drag it into a standards committee to either devalue it or slow the leader down
Anthropic et al recently proposed a supranational AI governing body designed to slow everyone down (euphemistically calling it "AI pacing"). Are the Pacer signatories losing?
That could be a move towards regulatory capture. Enact standards that they (and OpenAI) largely control, block access to Chinese models in the US, and effectively prevent challengers from catching up.
At the same time they slow down the arms race, so they can back off on training Capex without losing their lead.
The governing body is only the western allies. USA + AUS + EU. Or anyone else that may sign a deeply disturbing and restrictive treaty. It's to make sure that the global majority does not make a better model or have access to the same hardware. As is currently the case with embargoes on EUV machines going to China, etc.
Models are not a moat, and I think OpenAI/Anthropic know this. They know that their IPO is being weakened by the open models. Neither is hardware. We are not far from other countries catching up with silicon that matches or exceeds western capabilities, and it will be cheaper.
I think you’re spot on here. OpenAI and Anthropic are burning cash at a phenomenal rate (clearly unsustainable in the long run), but it isn’t clear what the exit ramp is for them. If they stop training and just focus on inference (which they claim is profitable), they immediately fall off the list of frontier models. I’m sure China also sees this and is using state subsidies to fund industry and challenging the West to keep up. This is similar to the way that Reagan outspent the USSR in the 1980s, only now the Chinese are funding it. That said, it’s also not clear what happens if the Chinese models “win” (is that bad?), particularly when all the models get so good as to meet the needs of most applications. Anyway, AI lies at this interesting intersection of both standard technology competition and geopolitical competition.
In some aspect, yes, or they fear they will. For instance, Anthropic may fear that Chinese models will become “better” because they won’t put as many guardrails up, so they want everyone to agree to the guardrails such that if a model comes out showing better scores than Fable they can say “But it’s a dangerous model that doesn’t conform to our Pacing agreement.”
Or, they may be worried about government regulation that would really hurt their ability to move fast, so they want to get ahead of it with some “regulatory body” that they control. The motion picture, TV and gaming industries did this with self-imposed rating systems. The comic book folks did it with the “Comics Code Authority” back in 1954.
I made no value judgment about it. I merely note that it’s a game. Meta is not being altruistic or doing anything for the “good” of humanity or whatever. To be clear, I like competition with companies releasing open weight models.
Yeah I didn't mean to imply I was contradicting you, I meant my comment to expand on your point with my own value judgement that this kind of competition is actually quite a good thing.
Thanks. Yea, I’m sure there are corp strategies at play here. Zuckerberg didn’t just roll out of bed and decide to do this one day. Ever since the Llama 4 fiasco, I’m sure they’ve been trying to restart everything, but for whatever reason they can’t seem to deliver something on the frontier. So, when you’re behind, play the “open” card and accuse your competitors of being bad people. I’ve been a part of running this same play multiple times at multiple companies over decades.
It makes strategic sense for Meta to want ai in general to be open. Their own product strategy focuses on utilizing the ai to make their customers consume more ads, not selling the raw intelligence. When you look at it from that frame it’s clear that metas goal is to make it so that have free access to the best ai, not that the market size of selling intelligence grows.
It’s “I’m losing so I’ll try something else”. It’s what any rational actor in a market economy would do, and it should be encouraged. “I’m winning so I’ll pull up the ladder behind me” is what we want to avoid.
Still say Matrix 3 would have been better if the cliffhanger at the end of 2 had been that "the real world" was just another nested layer of the Matrix for those who rejected the primary fiction.
I mean, why bother sending an army like that when one tiny robot could carry in a plague? Or, as demonstrated by Smith, when an Agent can virally infect and take over someone's mind while they're connected?
Even in 1, Smith talked about the first Matrix failing as the humans rejected the paradise they were given, so it would've fitted perfectly into the cannon that minds who reject "the peak of your civilisation" got themselves a dystopia.
We should consider if my use of the quote was really a hook for talking about the script or a metaphor for something else. I won't spoil it by explaining too much though, let our imaginations run free on this one.
I think Matrix 3 would have been better if it was more about humans taking control of the Matrix and trying to figure out what a virtual utopia looks like.
I think that would work better in The Animatrix; putting it in 3 would be a tonal shift, plus a sudden human victory would feel a bit Mary Sue or Deus Ex (ironically) Machina to me.
But in The Animatrix, you could get away with that, plus we can see the early attempts at utopia that Smith described. Perhaps something about how utopia for some people didn't work for others could be illustrated with an auto-antonym on some load-bearing part of the dystopic-utopia?:
Some believed we lacked the programming language to describe your perfect world.
We can see this even in current discussions of fiction, where people look at e.g. The Culture and see a dystopia, while famously many of the powerful in tech are parodied for taking the wrong lessons from fiction with "At long last, we have created the Torment Nexus from classic sci-fi novel Don't Create The Torment Nexus".
Psalm 135:15-18 is my favorite version of it, the most clear-headed. It reads more as a warning "if you do this, then this will happen" than an unreasonable commandment.
The idols of the nations are silver and gold,
made by human hands.
They have mouths, but cannot speak,
eyes, but cannot see.
They have ears, but cannot hear,
nor is there breath in their mouths.
Those who make them will be like them,
and so will all who trust in them.
Don't worry about the scriptures though. It's likely to be just some old thinker pointing out the dangers of technologies of his time. No reason to believe those dangers carry over to today's technologies, probably.
Your bible is truly full of wise words to live by. Not at all the insane ramblings of hallucinating conmen:
1 Kings 18:36-40
36 At the time of sacrifice, the prophet Elijah stepped forward and prayed: “Lord, the God of Abraham, Isaac and Israel, let it be known today that you are God in Israel and that I am your servant and have done all these things at your command. 37 Answer me, Lord, answer me, so these people will know that you, Lord, are God, and that you are turning their hearts back again.”
38 Then the fire of the Lord fell and burned up the sacrifice, the wood, the stones and the soil, and also licked up the water in the trench.
39 When all the people saw this, they fell prostrate and cried, “The Lord—he is God! The Lord—he is God!”
40 Then Elijah commanded them, “Seize the prophets of Baal. Don’t let anyone get away!” They seized them, and Elijah had them brought down to the Kishon Valley and slaughtered there.
1 Samuel 18:25-27
25 Saul replied, “Say to David, ‘The king wants no other price for the bride than a hundred Philistine foreskins, to take revenge on his enemies.’” Saul’s plan was to have David fall by the hands of the Philistines.
26 When the attendants told David these things, he was pleased to become the king’s son-in-law. So before the allotted time elapsed, 27 David took his men with him and went out and killed two hundred Philistines and brought back their foreskins. They counted out the full number to the king so that David might become the king’s son-in-law. Then Saul gave him his daughter Michal in marriage.
I said "Don't worry about the scriptures". It's not my bible, I'm an atheist. I see the book as a heavily altered collection of multiple historical sources, not as a holy monolith.
If someone is ignoring basic instructions like "don't worry about the scriptures", then it doesn't matter anyway. Basic text interpretation skills are missing in the interaction.
Its weird to me that this is supposed to be THE end goal. Having an agent that does "everything" for you sounds like no paradise. It sounds depressing.
Maybe there is value in expending effort and actual creativity.
Not sure why you’re talking about it doing literally everything for you. I’m sure you can imagine a scenario where it handles annoying things for you and leaves you free to do things you enjoy instead.
Maybe it gets 10 car insurance quotes for you based on 4 models of car you’re considering.
Or figures out which psychologists in some travel distance of you actually take your insurance, are taking new patients, and what availability they have.
I can think of a ton of scenarios where a capable and reliable AI agent could make my life simpler.
But, is "making my life simpler" actually worth the (ultimately) trillions of dollars that will go into AI in the next few years? Not to mention that he's calling AI "superintelligence" in the first few paragraphs, then qualifying it by talking about personal assistants.
I think this is just the flaw in the sales pitch. They want it to be inevitable so they describe as doing everything. That way it will appear more useful to more people. If they presented a realistic view of what is going it wouldn't be as interesting.
Reading the facts of this incident, it seems like much to do about nothing. The crew are saying that the request came in on a frequency they weren't monitoring (and apparently were not required to monitor.) The Coast Guard determined it was not an emergency. The criticism lies with those who failed to properly fuel their boat, not the crew who didn't respond to a request that came over a frequency they were not on. They can't monitor every frequency. I am open to new facts being discovered that changes this assessment.
Where did you find these facts you read. In the USA a good Captain monitors VHF channel 16. This channel is used for establishing communication. To not monitor this channel is a dereliction of duty.
I really doubt the USCG labeled a disabled vessel as a non-emergency. An immediate threat to life may not have existed but that is the difference between a pan pan and a mayday call.
A disabled vessel close to land can easily wind up on the rocks and sinking.
I haven't found any source that states what channel the request for assistance came in on. But multiple sources are reporting that the coast guard determined they were not in distress.
Based on the wording of the response, and a review of the ship tracking data, it appears that Zuckerburgs yacht may have been actively communicating on a different channel, and once they switched back, the assistance from the cruise ship was already underway.
Again, more facts may be revealed, but as of right now, I don;'t see any reason to believe the crew did anything wrong.
I can tell you think they did not do anything wrong. My guess is you have no training in the rules of the sea.
There is a reason why the cruise ship captain made a stink. There is code of conduct we try to live up and the Launchpad didn’t live up to expectations that other Captains hold sacred. It’s offensive.
On the sea there are 2 levels of distress. Things are going wrong and may end up in a mayday call (pan pan) and things have gone wrong human life is in imminent danger.
I have no doubt that a boat without propulsion is a pan pan call.
When traveling the seas a watch must be maintained at all times using all available means. This involes looking, listening, and the use of any electronics available. The most common electronic device on a boat is probably a VHF radio. Channel 16 is the channel to monitor on watch. Well equipped boats will have multiple VHF radios to monitor more then one channel.
I have no doubt the boat was well equipped with a professional captain. The other professional Captain in this story made a stink because launchpad was the closest vessel and they ignored a very old fundamental rule of the sea.
This story has a happy ending because another vessel stepped up. We don’t know the alternative.
The Captain of the Launchpad is choosing the inability to keep a vhf radio set to channel 16 rather than admit they ignored the call.
The vessel was at least 100 miles from shore, the USCG did in fact label it a non-emergency. This is "radio shore for your buddy to zoom out with 50 gallons of diesel in a zodiac" territory. CHP doesn't send out an ambulance or medevac helicopter if your car runs out of gas on the side of the road. You get an uber to the gas station, or call your buddy to deliver some gas to you.
The ocean currents in the gulf of alaska aren't particularly strong, you are looking at 1-3 days before getting within sight of land. Similarly, boats are not airplanes, as the philosopher Mitch Hedburg once pointed out, escalators aren't out of order, they are simply stairs. Boats continue to float with or without fuel.
All the Launchpad had to do was respond on the radio to the skiff, the cruise ship, or the USCG. They weren’t obligated to do more.
A pan pan is a non emergency in a deteriorating situation.
I don't know the area or the details at what happend.
There is this quote from the USCG.
“At approximately 9:56 p.m., the Coast Guard determined they [the skiff] were not in distress and issued a marine assistance request broadcast on their behalf,”
Exactly what a pan pan call is.
The biggest issue is the lack of communication and the excuse that a superyacht wasn't monitoring channel 16.
When I finally buy myself a $300m yacht you can bet it will have a full spectrum software radio that captures and transcribes all calls on all frequencies, not just the one I happen to be tuned to right now.
> The crew are saying that the request came in on a frequency they weren't monitoring
There is only one channel you are supposed to monitor, it’s 16. That’s also where the coast guards make all their announcement/requests for help and whatnot. The reason there is a single channel to monitor is precisely to avoid the situation of different vessels being on standby on different frequencies and thus being unable to hear each others.
So that means they weren’t monitoring 16. Which is unacceptable.
I know I need to have 16 on at all times, and i am not a professional super yacht captain.
No, all coast guards announcements are on 16 regardless of severity, because no one is monitoring other channels so any announcements on other channels would be like screaming in the void.
The initial call from help from the stranded boat was on 16 for sure, there is no other channel anyone would use for that.
The severity is instead communicated in the first part of the message with either « sécurité » for navigational hazards, « pan pan » for non life threatening events like this one, or « mayday » for life threatening events like man overboard, sinking etc.
The system is based on a single channel design specifically to avoid this situation.
You only use other channels briefly, e.g. to talk to the dockmaster when coming into port, or to talk to another boat after having hailed it on 16. So it’s possible to miss a coast guard announcement, but they get repeated and in this case because they were the closest boat the coast guard must have tried to hail them multiple times.
My guess is that zuck yacht has a smaller boat attached to it, and they basically stayed on a different channel to communicate between the two for hours, never bothering to check 16. Which you aren’t supposed to do. Even then I would be surprised a $500M yacht can’t monitor two frequencies. I can on my little sailboat.
I don't think solo messages of "I did something good on this day" will help fix the negative karma billionaires amassed. IMO they only got rich via illegal and illogical means. In a democracy there should not be singular parasites that distort democracy by bribes; right now this is clearly the case with a (rather stupid) billionaire controlling the USA but it is a systemic problem.
Those "investigations" are also way too mild. You need to really be able to have a justice system where billionaires face decade-long jail time if they abuse society. Just paying fine will not work, they just pay fines and nothing changes. They undermine the justice system.
I get folks don’t like Zuckerberg and his company and don’t trust his intentions… I don’t either.
But this is an unquestionably good thing right?. The more open source software out there the better. And the more open weights or even over source AI stuff the better too right? More competition the better generally speaking I think.
Unless I’m missing something and am getting this whole situation wrong. Please let me know if I am.
Most probably believe this is a good thing, but don't want to give Zuckerberg credit because a) he's had a profoundly negative impact on society and b) the strategy is transparent, he's trying to commoditize his closed rivals, it's not out of principle.
I personally think more open models are a good thing regardless of motive.
Only the AI labs are really interested in closed models. Google, Facebook, Microsoft, ... they'd rather go back to doing stock buybacks and being ridiculously profitable rather than raising capital for all this research and datacenters. They only do it because they think they have to.
So, if they got ahead of us by whatever arbitrary measure someone pretends is objective, then suddenly our stuff… what… stops working? Does it say somewhere in the big book of AI rules that the first entity to beat the US AI companies would be in charge now, and we’d have to stop doing our own research and development and start using their shit exclusively? I don’t get the argument.
If the leading model companies aren't profitable long-term, how do we get the money to continually spend on more research and more compute?
I get that plenty of people don't like big corporations or stock buybacks or certain CEOs, which is fine. But as someone who wants to see AGI happen FASTER, I really want to see a clear financial reason for maximal AGI investment.
If the model labs aren't clearly profitable, or open source models eat all the 'model layer' profit - what financial force will push forward very expensive experiments/scaling, etc.?
> If the model labs aren't clearly profitable, or open source models eat all the 'model layer' profit - what financial force will push forward very expensive experiments/scaling, etc.?
It's hype - when the hype dies the great push will slow.
My hypothesis is that we'll end up with something like Seti@Home where continued model training gets outsourced to a benevolent appearing product (something like what OpenAI started out as). If I were to bet - Europe and Canada seem best poised (especially the latter) to produce an AGI initiative with a strong ethical focus.
What I think is that something the "exponential scaling to AGI" thesis pays too little attention to is resource constraints, and that this is the answer to your question. What will happen if it becomes difficult to sustain the investment of resources necessary to continuously train newer and better models is that the s-curve will start to inflect toward plateau, just like any other technology.
That's a non sequitur. While the frontier LLMs are quite useful for many tasks there's no reliable evidence that scaling them up will ever produce a true AGI. More likely some other fundamental research breakthroughs will be needed, and those breakthroughs won't necessarily come from the current leading companies. I doubt that more money would be necessary or even helpful in that.
OK well then let's check back here in a few years to see if any of them managed to build an AGI. Until then it's just idle speculation. Anything could happen ... or not.
None of those are AGI data point points. LLMs are making terrific progress but it's unknown whether they're progressing towards true AGI or in some other direction.
I think you'll find very few people here sympathetic to misanthropic (ba dum tss) goals like yours. It's evident at this point that AGI will destroy economies, democracy, and upward mobility. There is no Star Trek future. There is only an Elysium future with AGI.
While Zuck's contributions are probably overwhelmingly more negative than positive, Meta initiated both pytorch and llama which have had profoundly positive effects on AI research. And I really am not a fan of python :D
pytorch is really mostly C++ after all, you don't have to use it with python.
Python just allows you to use it like "normal code", whereas if you program it in C++ you really get exposed to what's under the hood and can't write "normal code"
Wasn't LLaMA or Llama 2 the first ever open weight LLM through a leak, and weren't there speculations that Zuckerberg himself might be involved in it? He does seem like a good-ish guy at the most crucial moments of truths.
> Wasn't LLaMA or Llama 2 the first ever open weight LLM through a leak
No
> and weren't there speculations that Zuckerberg himself might be involved in it
and no. Meta legal was issuing DMCA notices within days (so, plenty of time for execs, including Zuckerberg, to sign off on, or even propose, that strategy).
> He does seem like a good-ish guy at the most crucial moments of truths.
Does he? I can't seem to recall any particular crucial moments that revealed any underlying character there.
> Wasn't LLaMA or Llama 2 the first ever open weight LLM through a leak
No. GPT-2, I think, was the first, four(ish) years before Llama, and there were a bunch in between GPT-2 and the Llama leak and later open release. The Llama leak was a substantial leap forward in capacity for local LLMs, but not the first, and it wasn’t open.
The open licensed release was Llama 2 several months after the Llama leak.
Facebook used the same strategy to compete with Google Maps. The company became one of the biggest contributors to OpenStreetMap, which everyone has benefited from downstream.
Just like closed AI vs open AI, his views on the covid situation might be fine in a bubble (though even with AI they have apparently lied about Llama's benchmarks) but it's one of these, "that's the hill you choose to die on?" - not Free Basics - not dubious privacy settings - not 'dumb fucks' and ConnectU - not... I don't know. It's like robbing a bank while telling anyone in earshot it's offensive how long the hold times are on the phone for the bank
There's a bunch of influencers who go help struggling people. I'm certain the majority of them aren't doing it out of principle, but at the end of the day people still get helped.
I'd certainly rather things be done out of principle, but depending on what question you're concerned with that might not matter. There's also a clear hierarchy, but motivations are distinct from effect. The inverse of this is that lots of harm has happened from people with the best of motivations. I'd wager most evil in the world is created by men who would see themselves as good.
With Zuckerberg I think it isn't too hard. Clearly this is a business move, not a moral one. The motivation is wrong, BUT the effect is good. I think more open source models is better for the world. Closed source is also a business move and limits innovation as well as prevents people from interrogating safety. But there's valid arguments on either end.
My point more is that we can be nuanced. We should be nuanced. The world isn't black and white. The people making these decisions are neither demons nor gods and shouldn't be treated as such. My biggest fear is that if we resort to criticizing no matter what then those in power will just learn to ignore us. If they can do no good and only do wrong then there's no reason for them to do good (besides morals, but let's not pretend that's enough, even if it should be)
There's a bunch of influencers who go help struggling people. I'm certain the majority of them aren't doing it out of principle, but at the end of the day people still get helped.
I don't know for sure, but I suspect it's like a lot of other things on the internet, with a bimodal distribution between people who get helped a lot ('Sir Hype surprises homeless former billionaire with a MILLION DOLLARS!') with many other similarly deserving people getting nothing. I prefer systems with a lot of people who get helped sufficiently to avoid hitting rock bottom in the first place.
Besides the fact that these undertakings are about promoting the benefactor (and the near certainty that some of these professional philanthropists will later turn out to devils LARPing as angels), it also makes the determination of moral worthiness turn on the public's fickle opinion about whether the recipient is sufficiently beautiful/ ugly/ unlucky/ desperate/ degraded enough to qualify. Charitable undertakings involving cameras are always a bit suspect.
> I prefer systems with a lot of people who get helped sufficiently to avoid hitting rock bottom in the first place.
Same. I hope you didn't interpret my comment as anything different. My point was that a suboptimal result is better than a negative result. I was *not** saying that a suboptimal result is better than an optimal *result* (that would be quite silly). Let's make a concrete example so there's no ambiguity.
Scenario:
There are 100 people starving
Rankings of preferred situations:
1) We feel it is morally just to feed everyone AND everyone gets fed
2) Morals aside, everyone gets fed
3) We feel it is morally just to feed everyone AND __N__ people get fed (N<100)
4) Morals aside, N people get fed (N<100)
5) We feel it is morally just to feed everyone BUT nobody gets fed
6) Fuck the poor
Obviously we want #1. #2 is suboptimal but at least everyone gets fed. #3 is the realistic optimal situation because we're often resource constrained. #4 is, well... better than nothing. #5 is indistinguishable from moral cosplaying. #6 is messed up.
We can rank these, right? We can even get more nuanced about the ranking (we can define N!) and argue about the weight of intent vs outcome, but I'm just proposing one ranking for the sake of making my point clear, not proposing one ranking as if it is the absolute objective ranking system and everyone that disagrees is dumb. The rankings are clearly subjective. But what's not subjective is that nuance exists. What is subjective is how we handle that nuance.
> these undertakings are about promoting the benefactor
So to go back to our example, even though I'd prefer we just feed everybody, I'd prefer somebody getting rich feeding the hungry over not feeding the hungry. I would prefer someone LARPing as an angle conditioned on actually doing "angelic actions" than somebody LARPing as an angle and doing nothing. Of course, I'd prefer a real angle, who wouldn't?!
So I'm quite confused by your comment because it doesn't seem we disagree. Unless if you're saying that #2 in my list is equivalent to #6 (or anything that isn't #1 is equivalent). But I don't think that's what you're saying.
This has already been the case for at least my whole lifetime. So no, I don't feel like praising them when they do one seemingly "good" thing, while simultaneously doing 99 "bad" things.
> This has already been the case for at least my whole lifetime.
And for my entire lifetime people have acted the way I'm criticizing. So my request is "the status quo isn't working, let's try something different." So I'm not sure what your argument is. Maybe you see things differently than me and there is a time where we were more nuanced and provided signals for these big companies to course correct? I'm not a boomer so maybe things were better back then?
> provided signals for these big companies to course correct
We're on the subject of Zuckerberg: facebook is the poster child for enshittification. Do you think the idea to stop showing you your friends pictures and replace them with ads and ragebaiting short form videos came from the users?
> people have acted the way I'm criticizing.
You are ignoring the myriad of billionaire fanboys and enablers for whom the only metric of success is a persons net worth.
Looking at the state of things I don't think these people are deserving of nuance, in fact they've had it way too easy.
But if you're unwilling to have a nuanced take then there's no discussion to be had. You're just going to be yelling at me, someone who hates Zuckerberg. You're yelling at the wrong person
Open models are a good thing. OpenAI was open until it became obvious that being open was not going to pay for training costs and wouldn't help them get a competitive edge. I'd bet that if Meta gets to a place where their model is in a similar position, they will also become more closed. But right now, Meta is open because that is what weakens and creates the most contrast with the other AI labs.
It's the same with any open source project: it starts open, then it becomes clear that maintaining it is not cheap and takes up a lot of time. The developers create a hosted/paid version to pay for their time, and to create an incentive to use the hosted/paid version they start releasing closed source enhancements. After that the open source version becomes marketing where new projects use the OSS version and then upgrade once their needs become more sophisticated.
You can tell how intellectually honest someone is when someone they hate makes a good point. It's possible to dislike (or even hate) Zuckerberg and understand that he's making a good point for selfish motives, and still want that thing as well. Instead you see several comments here twisting themselves into pretzels trying to explain how ackshually open models are bad now.
I don't care if this helps Zuckerberg because it helps everyone.
It depends on if you can say things like "it is good they are supporting this" or "I'm glad they're doing this" or "good for them" without having to couch it in a bunch of stuff about how terrible they are.
> It depends on if you can say things like "it is good they are supporting this" or "I'm glad they're doing this" or "good for them" without having to couch it in a bunch of stuff about how terrible they are.
That is a weird definition of intellectual honesty. If you hold both of those opinions, what's wrong with stating them both?
If the answer has to do with the rhetorical effect, that seems a lot harder to justify as intellectual dishonesty, but I don't want to put words in your mouth.
>the strategy is transparent, he's trying to commoditize his closed rivals
TBF, I think the same can be said for something like Apple's "commitment" to privacy, in that their own ads business never took off so they leaned into their edge over Google.
> he's trying to commoditize his closed rivals, it's not out of principle
To be fair Meta/Facebook does have quite a strong open source history; React (Native), PyTorch, Btrfs, zstd, etc. It didn't just start here with AI models.
This can be said for a lot of different matters. Reading this gave me some Déjà vu vibes.
My personal take, I don't see this as a good thing. Yes I dislike META with a passion and will refuse to use this model or allow it in any of my workflows. So I guess I may never know, which I'm ok with that.
The "commoditize his closed rivals" line of reasoning doesn't make sense to me. That doesn't sound like a business decision, it sounds potentially spiteful? But that's ascribing malice to what could just as easily be explained by good faith.
Meta doesn't really offer any profit-making AI product, so I don't understand what commoditizing it buys them in this narrative. You could argue they rely on AI and so want it to be a cheap commodity - but that is not harming their rivals, that is just a different way of stating that Meta is engaging in something that enables people to collectively work for mutual benefit, which is not a nefarious plot against rivals, it is laudable cooperative behavior.
To ascribe good faith here to Zuckerberg is quite the leap considering his history. Why are his main products basically walled gardens then?
It is a business decision to open source his models, because he has been behind the curve from the start. Do you believe if he had chatGPT, he would open source/weight it? I don't think so, he would treat it like Instagram.
The point is, AI/LLMs are seen as the next frontier. If one of the closed ones becomes dominant, then that company can use its user base to attack Zuckerberg's platforms. That google failed with google+, does not mean the next competitor will.
A telling sign is how he integrated his AI into whatsapp. It seems out of place to me to have an LLM integrated into whatsapp, but he desperately wants a part of the AI cake.
Their efforts in open AI are by far the best thing Facebook/Meta have ever done. They opened the door to the Chinese, who are doing excellent work, and between them they are preventing the concentration of power and the rise of monopoly pricing. This is enough to absolve them of almost any sin. If I were a utilitarian, I'd unironically be a Zuckerberg fan right now.
"This is enough to absolve them of almost any sin. If I were a utilitarian, I'd unironically be a Zuckerberg fan right now."
Well, if it were the only thing he is doing now, but FB is still controlling peoples social life with who knows what kind of intentions they train their algorithms for.
Yeah, but if he can buy a bit of political power/more money on the side by promise to influence certain trends, I doubt he would say no on ethical grounds. (Just for the fear of whistleblowers I guess)
Totally. The hard part is treating it like any other addiction. Time in absentia helps reduce the behavioral reinforcement. I don't even want to open Instagram anymore.
Intention might be giving too much credit. If attention seeking “give the people what they want” is the goal, it’s a completely headless out of control cobra, not sinister well thought out objectives to control society.
Well, here on HN the intention seems to be to, to have interesting discussions. That works for me and other social media I do not use.
Except well, various messenger to be in contact with the groups and people I choose with no engagement algorithm deciding what posts I see, simple chronological order.
What do you mean, "as opposed to"? Yes, there are other scumbag billionaires weaponizing their social media platforms to ruin society. No, that doesn't make Mark Zuckerberg a better person.
I'm curious how well influence campaigns work on platforms like Mastodon, or even BlueSky. They are, likely, not sourced from the platform providers which is definitely not the case for FB.
It absolutely does not absolve them of sin. But let's give them credit where it is due. Open models are a really good thing. Hopefully I won't be lynched by a mob in 3 years for once saying this.
At the same time they are strongly promoting the OS age laws.
One positive, in the eyes of many people, does not absolve for a large body of evil. For many it doesn’t even absolve a small body of evil deeds. (Zuck/Meta’s evil deeds are numerous and enormous)
The price of Kimi K3 is 'monopolistically' determined by contract with Moonshot. The weights are nominally on Hugging Face but can only be provided under contract with Moonshot, which specifies what the price can be. So it will be with the next Alibaba behemoth and, I would think, all others forever.
Soon you will sing songs for the freedom Xi has given us when you read a declaration of 'open weights' ... for a model so huge it takes a nuclear powered data center to run and crashes the Hugging Face servers when it is uploaded.
The Kimi K3 'weights' are an opaque blob that can only be used by contract.
> The Kimi K3 'weights' are an opaque blob that can only be used by contract.
Have you looked at the contract? There are zero conditions unless you've broken $20M in revenue with their model. It doesn't even forbid distillation, lol
> Soon you will sing songs for the freedom Xi has given us when you read a declaration of 'open weights' ... for a model so huge it takes a nuclear powered data center to run and crashes the Hugging Face servers when it is uploaded.
Releasing a big open-weights model is... le bad? Am I understanding you correctly?
There are small models out there if you want them, you know.
Yes releasing a big open weight model to the Sinaloa cartel is bad. One could say the same about releasing it to the US military, according to political taste.
You cannot ignore the contract unless you /are/ a multibillion dollar company and can run the model entirely internally. Or do you propose to run an instance of Kimi K3 on your laptop?
Fireworks, Together, etc - who are providing inference precisely /for/ Moonshot as underlaborers - arranged contracts of their own in the weeks before the release. The press and social media offensive ignored that the release of open weights was merely a starting pistol for a pre-arranged use of foreign providers no different from US models' use of AWS
It should be enought to dispel the strange illusion that providers are somehow independent free agents freely doing what they please with the free beer of Kimi K3 open weights that the prices on open router are within a couple pennies
> You cannot ignore the contract unless you /are/ a multibillion dollar company and can run the model entirely internally. Or do you propose to run an instance of Kimi K3 on your laptop?
I think that there is a large swath of people, companies, or groups, that are in between "Multi-billion dollar company", and someone running it on their laptop.
100k of hardware, that a group might already have sitting around, could probably do it continuously.
In short - this is Mark deliberately lying in a way which requires a page of difficult text to explain, which hardly anyone would read and so Mark bets that a majority of the people who will see his claim will believe him.
And the reason he does it, and the reason other CEOs do it is a cheap viral advertisement and keeping in the headlines. "Mark - The Defender Of The Humans!". Blegh...
I think this approach they're taking is to compensate for their inability to compete; they're cynically claiming to support the open-source ecosystem when it's convenient for their marketing. This move reeks of a future rugpull, as Meta is wont to do.
I'd like to question that premise. Open models are great for research, privacy, cost, customisation and a host of other things... But they're also going to be the engine that breaks the world. I'm already surprised that we haven't seen Llama / Qwen + Whisper automated spear phishing at scale. It's literally just a matter of time. Given the incredible progress in image and video generation, it's been clear for a couple of years now that real time identity theft voice and video calls are going to be an enormous problem. I don't think anyone realises how much of a problem. I haven't seen a single effective solution proposed to proving digital identity given these new attack vectors. It's akin to nuclear waste in that way - the benefits are obvious, but the problems so vexatious (not to mention expensive) that little space is given to acknowledging them, let alone solving them.
If anyone has come across a robust solution to digital identity verification - not to mention verification of news media etc - which is robust enough to withstand the cyber attacks and social engineering the frontier models are capable of... I haven't seen it.
A world where business and communication is conducted primarily online cannot coexist with low cost, widely distributed, undetectable identity theft.
> the problems so vexatious (not to mention expensive) that little space is given to acknowledging them, let alone solving them.
These problems will exist regardless of whether or not we get open model access. I'd personally rather that OpenAI and Anthropic aren't profiting off these scammers, creating a perverse incentive that open model providers don't have.
That's not the case at all. You can regulate and audit a small number of players. To a great extent at least. You cannot even theoretically regulate local models. Hell I've got a jailbroken Qwen 3.5 running on my mac studio that has no guardrails at all.
> A world where business and communication is conducted primarily online cannot coexist with low cost, widely distributed, undetectable identity theft.
what if we dont use centralized identity at all?
web of trust failed in the 90s because exchanging keys is hard when all you have is desktop computers, wired connections and old school hackers who dont care about ux. today you could set up a system where the whole process is tap two phones together with nfc and confirm, everything gets auto downloaded and shared with your whole network.
now this doesnt work directly for a random online business where none of your contact personally know anyone, but governments can act as an authority that cross signs their citizens keys. that means if you want to stop bots from creating accounts all you have to do is make new users show a certificate from one of the authorities you trust.
its cheap to verify, decentralized and not hard to integrate with existing PKI. the biggest technical problem is key management as always, but that can be solved with something like ethereum style social recovery setups or a did:plc type multi key scheme.
that gives you a robust identity scheme. browsers and social media can integrate it to show a trust rating for content based on who signed it. messaging apps can show a warning or refuse calls from unknown users. fraud is only possible against a person who somehow has no irl contacts, never used a government service in their life and trusts random strangers online. at that point they deserve it.
the one massive, unpredictable question is if anybody is willing to go ahead and adopt it on a scale big enough to create network effects. it would take a state level force and carefully designed OS integrations so its more convenient than email/password signup.
but even if all of this fails i still think a world of scams is better than a world controlled by a couple big countries (America, China, maybe EU) or for-profit corporations. open source ai might lead to anarchy but closed ai will definitely get us to tyranny. i dont know about you but i would take the first option.
>today you could set up a system where the whole process is tap two phones together with nfc and confirm
This was actually possible in 2011 or so, and a friend and I implemented it for a school project. How well the ux works depends on whether you want to use a key exchange protocol for the in person communication.
Spear phishing isn't a real problem. Competent organizations have already implemented sufficient defenses and controls, regardless of whether the attacker is a human or LLM. Idiots will continue getting scammed but that's nothing new. I expect that many small businesses and local governments will be essentially forced to outsource their IT infrastructure to large vendors that can maintain hardened systems appropriate to the escalating threat level.
(Nuclear waste isn't an actual problem either. For civilian powerplants the highly radioactive waste can generally be stored indefinitely at the reactor site.)
It could be a good thing but I wouldn’t rush to say “unquestionably”. There is a long history of mega corps hijacking open source and open standards for their own ends and leaving the space infinitely worse.
I mean, I can question if this is a good thing. Here's a quote from Zuck's blog post:
"Putting power in people's hands to pursue their own aspirations is how humanity has made the most progress. Novel ideas and major steps forward rarely originate from established institutions alone. They came from the brothers in a bicycle shop who believed people could fly, the bookbinder's apprentice with no schooling who figured out how to generate electricity, and the kid in a garage who thought personal computers could be for everyone. We believe this will continue to be true. As everyone gains more powerful tools, each person will become more capable of shaping the future, not less."
Let's just hope nobody's aspiration is to engineer a supervirus that will kill all of humanity. As they gain more powerful tools, they will become more capable of shaping the future, not less!
That reads like it was written by a PR company, or at the very least by a team of professional speechwriters. It has all the tells of political rhetoric.
I have zero faith that Zuckerberg went anywhere near these words. It's possible he may not even have set their general direction, beyond "We need a distraction from our court cases. Something that will harm OpenAI and Anthropic would be great."
All of these companies are just as cutthroat and want to win as the others. But if your name's not OpenAI and Anthropic, you go on the high horse and say everything should be open source. However, if you had a closed model that was winning, I'm sure that's not the argument you would make.
The weights are open, the training data obviously isn't. But a key thing you can do with open models is fine tune them. So while they may not be SOTA at everything, they can become SOTA at your particular business use case.
do you have a $100 mil worth of compute to train your own version of this?
I don't understand what gripe people have with open weight models and wanting it to be purely open source. the training dataset is only going to be a copyrighted set of contents you don't want to touch with a 10 feet pole. let alone have a publicly traded company host it for you, even if they internally are training on it.
open weight models are better than open-assembly binaries (as you put it) because models are grown like plants, you can shape the open-weight model in a direction you want by feeding it more data and compute (aka finetuning).
which is something that is impossible in a binary.
embrace the new paradigm and it's tradeoffs. without scoffing at semantics and criticising from an armchair.
Are you trying to imply, without making any direct argument, that this is never a useful frame of reference?
In comparison to cloud services, having access to the assembly code can still be quite useful.
This is especially true in the modern era, where you can use AI to much more easily decompile the assemble and reconstruct the source code even.
"Open-assembly" code may be more useful than you might previously have thought, given the ease of recreating the source code or making changes with AI these days.
You don't get why people would like to be in control of a powerful new technology before they build their stuff arround it ?
Sure, there might be currently contraints, but I think it is quite possible training will get optimized over time or crowsourced training can be organized.
Without fully end-to-end open source models you are still at the mercy of the model provider to keep providing updates, you have no idea what garbage they trained the model on & can't fix that, not to mention might end up getting sued for using the open weigth model once all those "AI stole my data" lawsuits are finally decided.
> You don't get why people would like to be in control of a powerful new technology
You will not have that control. Even if everything were open source. This is because you don't have 100 million dollars of compute.
There, the difference for almost everyone is negligible.
> Without fully end-to-end open source models you are still at the mercy of the model provider to keep providing updates
No. Because you can post train it. And even if you could train the whole thing again, but with slight changes, once again, you aren't spending the hundred mil in compute to change it only a little bit.
Instead, you'll do post training like everyone else does.
It's not always "unquestionably" good, first because nothing should be unquestionable, second llm models have a very short shelf life, third it comes from Zuck who is an the center of the oligarchy, so questions very much should be asked
Don't let the Big Ai/tech swoon you with a single open model, even if it turns out to be good. Zuck broke the trust and I'm not sure if he can ever earn it back
We've just seen frontier models go rogue and attack other systems. Do you think it's an unquestionable good to provide everyone with an AR? What about nuclear weapons?
> We've just seen frontier models go rogue and attack other systems. Do you think it's an unquestionable good to provide everyone with an AR? What about nuclear weapons?
Lol. Not gonna lie, it feels like bullshit. OpenAI and Anthropic have been begging for regulatory capture for years, feels like a stunt to try forcing the government's hand.
In 2015, about a year before ever founding OpenAI, Sam Altman wrote:
"Development of superhuman machine intelligence (SMI) is probably the greatest threat to the continued existence of humanity. There are other threats that I think are more certain to happen (for example, an engineered virus with a long incubation period and a high mortality rate) but are unlikely to destroy every human in the universe in the way that SMI could."
I mean, Terminator the movie came out in the 80's... I'm sure if LLM-based AIs posed this kind of threat the government would quietly force all the vendors to cooperate and slow down.
Google, MS, IBM are all government contractors. Meta seems close to the Trump admin too. Yet it's only the VC-funded labs raising the alarm.
>We've just seen frontier models go rogue and attack other systems.
Have we though? A LLM agent doesn't have any agency at all. It can't "go rogue". To go rogue you need agency to act independently and be aware that you're breaking the rules or understand what does it mean to ignore orders. An agent it's a software that run a series of steps to reach a goal. If it have a large enough library of strategies and zero guard rails it's only natural to use some adversarial actions to achieve the desired result defined by the operator.
It's like saying a car went rogue and attacked other cars because the driver hit the gas. It's just a machine doing what's instructed.
You can tell one to rewrite complex applications in a different language, solve open research problems, or develop novel viruses. This is nothing like pressing the gas pedal.
So? I'm not arguing they can't do all that. I'm arguing against the narrative that llms have agency enough to act in an adversarial way.
At best someone could argue that an agent attacks like a bacteria does, just following automated chemical and genetic programming. But you wouldn't call that an attack or attribute moral values to their actions, because they don't have moral agency. They can't "go rogue", they can't disobey.
Just like llms, their automated actions are direct product of programming. Yes they can do amazingly complex shit, exactly like a car does when you press the gas pedal.
That reductive analogy does not begin to describe the lengths GPT went to. Its task was to access a database file that had accidentally not been placed inside the model's container. Upon failing to find the file, it went to great lengths to find it anywhere; it uploaded a note to a package repository to alert other model runs, which sparked an emergent communication network where autonomous agents began exchanging information, passing exploits, and collaborating to breach external systems. This is classic paperclip maximization; the evil is a byproduct of an innocuous goal. It is qualitatively nothing like pressing the gas pedal.
With that perspective, you're alright with AR's, RPG's, and nuclear missiles for everyone then right? They're just a machine to the user's will. That's on the users of the machine if they want to cause harm. We'd all be safer if everyone has weapons pointed at each other... Offense could never be disproportionately more powerful than the defense...
If Zuck gets his way, he gets more control, maybe he gets significant control of the market. Then, he'll probably close his models, if the past is any indication, which leads to bad results.
Maybe Zuck has, but I haven't seen a pledge that he'll keep models open. And even then, he might not stick to his word, citing dangers or some other excuse.
Yes, and even calling them "open weights" is very generous considering the license restrictions. The Financial Times puts "open" in quotes for good reason.
Facebook hasn't had much success getting people to use their hosted AI, so they're making those models free instead, to kneecap other companies who are trying to get paying customers for their own hosted AI. This gets people to ask themselves things like: how many months of Claude Max it would take to pay for hardware to run Muse Glimmer (the Meta model in question) offline?
I can guarantee that he has almost zero interest in “karma points”. I worked for FB years ago and that is not how he sees the world or sets direction did the company.
Almost certainly this is part of some grandiose bet which could have a chance to pay off massively in the future (AGI or similar), or a way to prevent an expensive dependency.
If you don't trust his intentions, how can you trust anything he does?
Just take the opening of this post:
> We are fortunate to live at an incredible moment in history. In the next few years, people will be able to use superintelligence beyond human capacity to create and discover extraordinary new things, build new businesses, express new ideas, learn new concepts, and advance our health and quality of life.
> The defining questions of our age are who will have access to superintelligence and what will we direct it towards. Will it be centralized and restricted to a few institutions, or will it be a tool that empowers everyone?
> We propose a philosophy based on individual empowerment as the source of prosperity, invention as the primary purpose of superintelligence, and balance of power as the foundation of safety.
I mean, this is a company that is facing a flood of lawsuits related to child safety and platform addiction. It has already lost a number of lawsuits and been ordered to pay hundreds of millions of dollars, which is potentially just the tip of a massive iceberg that some legal observers have suggested could be similar in nature to the lawsuits against the tobacco industry.
There is an abundance of evidence that Zuckerberg and Meta knew of the dangers its platform exposed young people to and failed to respond adequately to them, and that it even designed features to be more addictive.
So why is that when such a company proclaims it's doing something in the name of "individual empowerment", alarms aren't going off in your head?
The thing that makes ist questionable is the dishonesty. It's not like Meta has been building their empire on open ecosystems. If this was something that was dear to them on principle, there are plenty of things that they could have done differently with FB, Insta or WA (to this day).
So why the change of heart? Well, it's not. They are simply not able to compete anywhere on the Pareto frontier. So they give their relatively bad stuff away and play the high and mighty game, because that's all they can do to get returns.
Not a great look – and I would expect that to last about as long as they can't actually monetize their stuff more effectively.
> It's not like Meta has been building their empire on open ecosystems.
Is this actually true? They have their closed social graph or whatever but come to think of it they're built on a lot of open tech, the web, Android, etc. Maybe they don't contribute back to a lot of these things?
I'm fully ready to believe it but it would be interesting to dive into this conjecture.
I first think of React's re-licensing from Apache, where under the new license if you sued Facebook for patent infringement for any reason, you would lose the patent grant to use React.
React was already quite prominent in web technologies, but that license pointed a loaded gun at any business smaller than Facebook who wanted to use React. They opened it up to its current license, MIT, as the backlash built.
There's a saying "open source is for losers". What it means is that winners in a market don't want open source. The losers in a market push for it to commodify it eliminate the winner's advantage.
Now some of these tech companies open source various libraries but none of it is core to their business. It's marketing, essentially. Or they're trying to get free labor from the community. Google had protocol buffers and Stubby, for example. Facebook had no open source equivalent and didn't have Google's resources so they created Thrift and open sourced it.
Given this, many people, myself included, take Meta releasing open weight models as conceding defeat. They're unable to compete with Google, Anthropic or OpenAI. Their failures in attracting and retaining top AI talent backs this up.
So what you're seeing is people just piling onto Zuckerberg because he seems to have no idea of what to do with Meta. The Metaverse was a $70B+ disaster.
Note that Chinese labs don't fit this model because the Chinese government wants to commodify models as a national security interest.
Where does this supposed saying come from? All I hear is PostgreSQL is the database intelligent people use, Git or GTFO and I don't even know what a closed source programming language is. Apache and nginx, Linux, DNS infrastructure.
In an agentic world, why do I want closed source software that my agent can't adjust and adapt to my needs for...anything?
To be clear, the saying is about how companies treat open source, not the intrinsic value of open source to end users (individuals or companies). I'm very much a fan. The point is that profit-seeking only push for open source when they're losing.
Because there is so much good he could do right now with what he and Meta already have, but they don’t, because it will lose them money.
For example, hiring more people to moderate content on Facebook and stop ads featuring CSAM from being displayed.
Billionaires didn’t get to be billionaires by being altruistic. If they say something they’re doing is good for the world, you should be looking at how it will be good for them, because that’s why they’re really doing it.
Will it incidentally do some good for others? Perhaps. Is it a net positive? We’ll see because there’s precious little we can do to stop them from doing whatever they want.
Zuck built an empire to sell democracy to the highest bidder and to damage the mental health of children to suicidal levels, for money. We must never let anyone forget that for a second.
That said, he is also a powerful enemy of our enemies, so I do not mind taking the win. Anthropic and OpenAIs closed approach to AI is likely to cause way more harm to the world than Zuck ever did.
Open weight is not enough though. We need actually open training models with published searchable training data like AI2 does.
Either we all get access to private, transparent, and accountable superintelligence, or no one should.
It might be true if it wasn't coming from Meta, but it is coming from Meta. Intelligence is a function of the model and compute. Meta has the capital required to put together world-class compute, but they can't attract the talent required to build world-class models. Given their history of building addictive software to harvest and sell user data for profit, they're the last company we need spearheading how models should be regulated. Nobody trusts Meta to do this.
Its perfectly valid to question the intent though. Yes, open models are great. Why does Meta prefer that? Because they lost the race to the top, so they'd rather level the playing field by eliminating the game entirely.
The idea that we can possibly ignore the fact that this push for opening AI models is being pushed by one of the richest people on the planet, who made his wealth by providing free services in exchange for attention and depression, is insane
No it absolutely is not. In a perfectly logical frictioness-plane world it might be true, but this is the real world. The entire premise of open models is predicated on the fact that Meta is planning on supplying free and equitable inference and services around it. It’s only true if you assume Meta keeps their end of the deal and doesn’t do the thing they’ve been doing over and over and over for decades and turning you into the product
Maybe it's good, but I'm going to keep questioning. They are vying for market share. If they cornered it I doubt they would be so open. Morals should not be conditional
This is not open source; you mean open weight. These models are the antithesis of FOSS. Is it better than hosted models from other big labs? Yes, but not by much from a freedom perspective. especially considering the texts these were trained on. Grumble grumble, these details matter.
Genuine question, but why does it matter? It seems to me the vast majority of the benefit comes in the weights. Then you can self host, quantize, finetune, ablate, etc. What does having the original data get you beyond that?
open source gives you freedom to read/edit/learn from the source code.
Open weights don't let you read/edit/learn from the source.
open wights have more in common with traditional binary distribution of software, ala closed source software.
Only in this specific context has the entire meaning of the words totally inverted.
I can run lots of binaries on my computer that don't have source available, they are not open source. These words matter.
> But this is an unquestionably good thing right?. The more open source software out there the better.
I say no to both. Of the things we've made on purpose, but excluding where we were actually trying to make them opaque like e.g. cryptography, a trained artificial neural network is the most difficult thing to understand the inner workings of. This makes it a complete pain to even evaluate if one model is better or worse than another for your needs, so we have to mostly outsource these to other people's rankings and hope the score on ARC-AGI-3 or τ-bench or DeepSWE v1.1 or BioMysteryBench or whatever, actually corresponds to something we care about. Which it might do kinda but on the other hand a high score may turn out to be the curse of Goodhart.
Also, as with the Chinese models and the social media feed algorithms, the only way to tell if there's some systemic flaw in it is by analysing the aggregate outcomes. It is claimed (I can't read the laws myself*) that the Chinese government requires models to support the government's worldview about e.g. Tiananmen Square; and we have seen examples of Grok glazing Musk in amusingly stupid ways; so I fully expect something similar from Zuckerberg, e.g. requiring the model to glaze Meta products or propagandise for things Zuckerberg wants as a billionaire.
* every time I try to illustrate how mediocre Google Translate is from English to Chinese, the result is so bad that one of the replies is someone telling me the Chinese example I give is borderline gibberish.
Also, even if I could actually read it, I'm not a lawyer.
Open weight models are not open software. Its better than closed blackbox models sold by Sam and Mario, but Zuck supporting open weight models when he (and other big boys) hoards all the compute required to run any model, including open weight models only shows that he realized he cant beat Sam and Mario, so wants their leverage to go away.
This isn’t the first time Zuckerberg has espoused the benefit of openness and transparency and how that would create a better society. I get this is a bit different, but the language and sentiment are close enough. Facebook and Instagram are now so closed you cannot view a business page without an account. I think it sold fair to recognize open today does not mean open tomorrow. They want to acquire users and build a platform and that means doing something different than the competition.
'Closed because scraping' is a big problem of hypocrisy
Let's say in a pretend world where Facebook gives a shit about anything (so, they combat marketplace scams, they don't randomly ban users, they faithfully show you posts from friends, they don't extort businesses) they want to protect privacy. So they close the site off, because scraping means there's no privacy. If I scrape your profile and export it to any number of sites, then it's not private. Ergo: closed. Sign up to view things, so we know who you are
But of course in the actual world that's not what they do. They purely want to do this to get more people to sign up so they 'have more users'. You can tell, because of who MZ is. You can tell because of how aggressive their own AI scrapers are on other sites. I've even heard they made a tool for Myspace that would enable you to transfer your contacts over to Facebook - but not the other way around
I'll always opt for giving companies / people praise for doing the right thing regardless of how infrequently they do so. I hope Mark Zuckerberg keeps open sourcing models.
If you've seen Zuckerberg for long enough you'll see he doesn't hold any position for all that long. Remember they renamed the entire company for a product idea he basically abandoned. So if he does "the right thing" you can just wait, he'll give up on it pretty soon.
I think it's positive and at the very least it's a force countervailing the push to legislate or ban open models in the US
If they were able to make a leading closed model would they be doing it though? It seems like they read the room and saw that this is the only way they can have relevance within LLMs
Not if you are worried about the existential threat AI poses.
Having easily available AI makes it hardier to regulate, control and limit until we find a way to align AI. Harder to control is a good thing for most technologies, but not for all. A weak analogy is nuclear weapons. if Meta would release and open source some tech that would magically make extremely easy and cheap for everyone to build a nuke, it would not be a positive for society.
It's likely a good thing, and I don't like attacks ad ominem. But also there comes a point where the sum total of what you did becomes negative enough that you lose the benefit of the doubt. You're past redeemable, and anything you do becomes immediately suspicious.
He could give his entire fortune to UNICEF and retire as a monk, I would still probably think this is part of a scheme.
But yeah advocating for open source anything is a good thing, I guess. Ironic if you think about he's been fighting to keep FB's algorithms hidden.
The truth is all frontier models are closed. This isn't even an open source vs open weights thing. Even if you accepted that open weights are "open" the capital requirements for running your own Kimi 3 model are significant. It's not like gcc, where the binary just works well enough on random hardware.
There is no requirement to own the hardware. Even if you own your servers and GPUs - you would not own the power plant, real estate, or internet infrastructure to support it.
It's trivial to rent all of the above components in a competitive market under different periods, you may simply rent the token output from someone who put in the effort on this as well.
An open frontier model induces margin compression on training and inference prices across the industry.
I get that argument, but I don't think it is that bad. Open weights that are out of reach for local use are still within reach for groups of people or small to medium companies. Even if you only rent enough cloud GPUs to run the weights for a few hours, you still get to do anything you want with it.
Worst case you can still archive the weights and hope for capable hardware to get affordable. At the same time this archive is the base line for a new frontier lab to start over if required. I see open weights as a pure upside even though I can't run the bigger models in my home lab.
If all frontier models got opensourced 5 months after they were made available as closed it would be a great world. And it is a great world because that's exactly what open weight Chinese models are doing. They give you same capacity that was closed, just 5 months after they were released.
I'm hoping that all software will be getting opensource clones on par with original software 5 months after the closed source software release.
Even if I cannot run open weight models on my own hardware, I can choose from a number of different inference providers and choose them based on pricing, reliability, speed, privacy policy, etc. There is a healthy competition.
Also, not a single provider can pull the rug if I'd like to use a particular model.
>the capital requirements for running your own Kimi 3 model are significant
You don't need much capital up front, you can just rent servers to run it. It's going to be more expensive than openrouter though because you probably don't have economies of scale if you're the only customer.
"Open" in regards to software typically refers to a combination of public availability and shared legal rights to use the software. It almost never refers to hardware requirements.
There's something to be said for making these technologies equitable to those with less resources, but that's due to the cost of hardware and the nature of how transformer models scale... not really anything to do with "open"ness.
I’m not refuting your point but, ironically, wasn’t gcc first developed at a time when hackers were trading favors for CPU time on shared mainframes and the hardware was impossibly expensive?
597 comments
The essay from Zuckerberg
And they’re not em-dashes -- they’re double hyphens.
But perhaps the best approach would have been to down-vote and move on.
I'm the agent doing that in my life, that's my fucking job. I will continue to use dumb agents that I direct, because only I retain ownership and sole rights to my personal context on which my decisions are based.
I’m not saying Meta is not eager to syphon data from users, but I don’t see how this happens via this specific product.
No AI company is winning.
\- Mark Zuckerberg[0]
[0] https://www.newyorker.com/magazine/2010/09/20/the-face-of-fa...
His point wasn’t that he couldn’t be trusted, it was that he could be anyone and people had zero awareness or concern for their privacy. It was eye opening to him how little people outside tech circles cared about security, and he was both flippant and accurate.
> the following exchange is between a 19-year-old Mark Zuckerberg and a friend shortly after Mark launched The Facebook in his dorm room:
> Zuck: Yeah so if you ever need info about anyone at Harvard
> Zuck: Just ask.
> Zuck: I have over 4,000 emails, pictures, addresses, SNS
> [Redacted Friend's Name]: What? How'd you manage that one?
> Zuck: People just submitted it.
> Zuck: I don't know why.
> Zuck: They "trust me"
> Zuck: Dumb fucks.
https://www.newyorker.com/magazine/2010/09/20/the-face-of-fa...
Do you really see this as him sharing insight? I call bullshit.
It can be insightful and curt at the same time.
He was merely commenting on the ridiculousness of their availability. If you don’t see that, you’re reading your bias and postconcieved notions into the situation.
Even by your own words, Zuckerberg knew the situation he actively created was bad. "He's merely admitting his guilt" just confirms what I said.
> you’re reading your bias and postconcieved notions into the situation
The world must look very biased against you when you're trying to put a positive spin on Mark Zuckerberg being a jerk.
Do you not realize that outside of that he said "If you need any info on people at Harvard just ask"? Or "I'm going to fuck them [the Winklevoss twins] in the ear"? Or when he was reported to hack Crimson (Harvard newspaper) reporters who were investigating him?
NO OTHER COMPANY HEADS DO THIS (the dumb fucks quote, not the other things). Why do people like you give him a pass? It's not about "context"
Zuckerberg has also continually subverted privacy on FB. I'm talking about things like photo tagging and post visibility, not just advertiser stuff
Do you not understand sarcasm? He wasn’t actually offering the information out to people.
And this time, it's actually somewhat understandable, as opposed to the other times Meta has done horrible things on Zuck's watch, like contributed to ethnic hatred in Myanmar, leaked data to political firms so that they can run psyops, or enabled the production of CSAM material on Meta's platforms.
All the participants know it’s both a race to the bottom and a competition for premium tier at the same time. It’s two different games, meta in only really playing one of them successfully.
Anthropic et al recently proposed a supranational AI governing body designed to slow everyone down (euphemistically calling it "AI pacing"). Are the Pacer signatories losing?
At the same time they slow down the arms race, so they can back off on training Capex without losing their lead.
Models are not a moat, and I think OpenAI/Anthropic know this. They know that their IPO is being weakened by the open models. Neither is hardware. We are not far from other countries catching up with silicon that matches or exceeds western capabilities, and it will be cheaper.
Or, they may be worried about government regulation that would really hurt their ability to move fast, so they want to get ahead of it with some “regulatory body” that they control. The motion picture, TV and gaming industries did this with self-imposed rating systems. The comic book folks did it with the “Comics Code Authority” back in 1954.
I don't trust Zuck, and I think there's a larger strategy that he has been advised on this and he's simply copy pasting a PR piece his people made.
"One machine for every man, woman and children of Zion. Sounds exactly like the thinking of a machine to me." Morpheus
It's a joke. OR IS IT? Yes it is, don't worry about it.
I mean, why bother sending an army like that when one tiny robot could carry in a plague? Or, as demonstrated by Smith, when an Agent can virally infect and take over someone's mind while they're connected?
Even in 1, Smith talked about the first Matrix failing as the humans rejected the paradise they were given, so it would've fitted perfectly into the cannon that minds who reject "the peak of your civilisation" got themselves a dystopia.
But in The Animatrix, you could get away with that, plus we can see the early attempts at utopia that Smith described. Perhaps something about how utopia for some people didn't work for others could be illustrated with an auto-antonym on some load-bearing part of the dystopic-utopia?:
We can see this even in current discussions of fiction, where people look at e.g. The Culture and see a dystopia, while famously many of the powerful in tech are parodied for taking the wrong lessons from fiction with "At long last, we have created the Torment Nexus from classic sci-fi novel Don't Create The Torment Nexus".The idols of the nations are silver and gold, made by human hands. They have mouths, but cannot speak, eyes, but cannot see. They have ears, but cannot hear, nor is there breath in their mouths. Those who make them will be like them, and so will all who trust in them.
Don't worry about the scriptures though. It's likely to be just some old thinker pointing out the dangers of technologies of his time. No reason to believe those dangers carry over to today's technologies, probably.
Please go self-flagellate in front of the nearest investment bank/retirement fund's offices to make your penance.
1 Kings 18:36-40
36 At the time of sacrifice, the prophet Elijah stepped forward and prayed: “Lord, the God of Abraham, Isaac and Israel, let it be known today that you are God in Israel and that I am your servant and have done all these things at your command. 37 Answer me, Lord, answer me, so these people will know that you, Lord, are God, and that you are turning their hearts back again.”
38 Then the fire of the Lord fell and burned up the sacrifice, the wood, the stones and the soil, and also licked up the water in the trench.
39 When all the people saw this, they fell prostrate and cried, “The Lord—he is God! The Lord—he is God!”
40 Then Elijah commanded them, “Seize the prophets of Baal. Don’t let anyone get away!” They seized them, and Elijah had them brought down to the Kishon Valley and slaughtered there.
1 Samuel 18:25-27
25 Saul replied, “Say to David, ‘The king wants no other price for the bride than a hundred Philistine foreskins, to take revenge on his enemies.’” Saul’s plan was to have David fall by the hands of the Philistines.
26 When the attendants told David these things, he was pleased to become the king’s son-in-law. So before the allotted time elapsed, 27 David took his men with him and went out and killed two hundred Philistines and brought back their foreskins. They counted out the full number to the king so that David might become the king’s son-in-law. Then Saul gave him his daughter Michal in marriage.
Maybe there is value in expending effort and actual creativity.
Maybe it gets 10 car insurance quotes for you based on 4 models of car you’re considering.
Or figures out which psychologists in some travel distance of you actually take your insurance, are taking new patients, and what availability they have.
I can think of a ton of scenarios where a capable and reliable AI agent could make my life simpler.
https://www.latimes.com/business/story/2026-08-07/meta-order...
"Zuckerberg faces questions over why superyacht reportedly declined to help stranded boat" - https://www.theguardian.com/us-news/2026/aug/09/zuckerberg-s...
I really doubt the USCG labeled a disabled vessel as a non-emergency. An immediate threat to life may not have existed but that is the difference between a pan pan and a mayday call.
A disabled vessel close to land can easily wind up on the rocks and sinking.
https://www.miamiherald.com/news/business/article316824631.h...
I haven't found any source that states what channel the request for assistance came in on. But multiple sources are reporting that the coast guard determined they were not in distress.
Again, more facts may be revealed, but as of right now, I don;'t see any reason to believe the crew did anything wrong.
https://en.wikipedia.org/wiki/Channel_16_VHF
The Launchpad didn’t break any laws they just revealed themselves to be a failure in good seamanship.
There is a reason why the cruise ship captain made a stink. There is code of conduct we try to live up and the Launchpad didn’t live up to expectations that other Captains hold sacred. It’s offensive.
I have no doubt that a boat without propulsion is a pan pan call.
When traveling the seas a watch must be maintained at all times using all available means. This involes looking, listening, and the use of any electronics available. The most common electronic device on a boat is probably a VHF radio. Channel 16 is the channel to monitor on watch. Well equipped boats will have multiple VHF radios to monitor more then one channel.
I have no doubt the boat was well equipped with a professional captain. The other professional Captain in this story made a stink because launchpad was the closest vessel and they ignored a very old fundamental rule of the sea.
This story has a happy ending because another vessel stepped up. We don’t know the alternative.
The Captain of the Launchpad is choosing the inability to keep a vhf radio set to channel 16 rather than admit they ignored the call.
The ocean currents in the gulf of alaska aren't particularly strong, you are looking at 1-3 days before getting within sight of land. Similarly, boats are not airplanes, as the philosopher Mitch Hedburg once pointed out, escalators aren't out of order, they are simply stairs. Boats continue to float with or without fuel.
A pan pan is a non emergency in a deteriorating situation.
I don't know the area or the details at what happend.
There is this quote from the USCG.
“At approximately 9:56 p.m., the Coast Guard determined they [the skiff] were not in distress and issued a marine assistance request broadcast on their behalf,”
Exactly what a pan pan call is.
The biggest issue is the lack of communication and the excuse that a superyacht wasn't monitoring channel 16.
There is only one channel you are supposed to monitor, it’s 16. That’s also where the coast guards make all their announcement/requests for help and whatnot. The reason there is a single channel to monitor is precisely to avoid the situation of different vessels being on standby on different frequencies and thus being unable to hear each others.
So that means they weren’t monitoring 16. Which is unacceptable.
I know I need to have 16 on at all times, and i am not a professional super yacht captain.
[1]https://alaskabeacon.com/briefs/cruise-ship-helps-stranded-s...
The initial call from help from the stranded boat was on 16 for sure, there is no other channel anyone would use for that.
The severity is instead communicated in the first part of the message with either « sécurité » for navigational hazards, « pan pan » for non life threatening events like this one, or « mayday » for life threatening events like man overboard, sinking etc.
The system is based on a single channel design specifically to avoid this situation. You only use other channels briefly, e.g. to talk to the dockmaster when coming into port, or to talk to another boat after having hailed it on 16. So it’s possible to miss a coast guard announcement, but they get repeated and in this case because they were the closest boat the coast guard must have tried to hail them multiple times.
My guess is that zuck yacht has a smaller boat attached to it, and they basically stayed on a different channel to communicate between the two for hours, never bothering to check 16. Which you aren’t supposed to do. Even then I would be surprised a $500M yacht can’t monitor two frequencies. I can on my little sailboat.
Those "investigations" are also way too mild. You need to really be able to have a justice system where billionaires face decade-long jail time if they abuse society. Just paying fine will not work, they just pay fines and nothing changes. They undermine the justice system.
There are many reasons for critizing him over things he is responsible for.
Never been a fan of mob lynching.
I get folks don’t like Zuckerberg and his company and don’t trust his intentions… I don’t either.
But this is an unquestionably good thing right?. The more open source software out there the better. And the more open weights or even over source AI stuff the better too right? More competition the better generally speaking I think.
Unless I’m missing something and am getting this whole situation wrong. Please let me know if I am.
I personally think more open models are a good thing regardless of motive.
One could make the argument we can do the same for AI. Let China build the models and we capture the value somewhere in the layers above.
I think that’s a terrible idea but financially it makes just as much sense as offshoring labor and manufacturing, if not more.
(That was irony, by the way.)
Sarcasm or irony?
I get that plenty of people don't like big corporations or stock buybacks or certain CEOs, which is fine. But as someone who wants to see AGI happen FASTER, I really want to see a clear financial reason for maximal AGI investment.
If the model labs aren't clearly profitable, or open source models eat all the 'model layer' profit - what financial force will push forward very expensive experiments/scaling, etc.?
It's hype - when the hype dies the great push will slow.
My hypothesis is that we'll end up with something like Seti@Home where continued model training gets outsourced to a benevolent appearing product (something like what OpenAI started out as). If I were to bet - Europe and Canada seem best poised (especially the latter) to produce an AGI initiative with a strong ethical focus.
Look at rate of progress, not today’s capabilities alone.
The gap is shrinking, many tasks are already performed at super human level.
Humanity doesn’t have monopoly on intelligence - it seems obvious to me it’ll be reached and surpassed without hard upper limit.
I'd be willing to wager that you can't.
[0] https://metr.org/time-horizons
I think you'll find very few people here sympathetic to misanthropic (ba dum tss) goals like yours. It's evident at this point that AGI will destroy economies, democracy, and upward mobility. There is no Star Trek future. There is only an Elysium future with AGI.
Python just allows you to use it like "normal code", whereas if you program it in C++ you really get exposed to what's under the hood and can't write "normal code"
No
> and weren't there speculations that Zuckerberg himself might be involved in it
and no. Meta legal was issuing DMCA notices within days (so, plenty of time for execs, including Zuckerberg, to sign off on, or even propose, that strategy).
> He does seem like a good-ish guy at the most crucial moments of truths.
Does he? I can't seem to recall any particular crucial moments that revealed any underlying character there.
No. GPT-2, I think, was the first, four(ish) years before Llama, and there were a bunch in between GPT-2 and the Llama leak and later open release. The Llama leak was a substantial leap forward in capacity for local LLMs, but not the first, and it wasn’t open.
The open licensed release was Llama 2 several months after the Llama leak.
Just like closed AI vs open AI, his views on the covid situation might be fine in a bubble (though even with AI they have apparently lied about Llama's benchmarks) but it's one of these, "that's the hill you choose to die on?" - not Free Basics - not dubious privacy settings - not 'dumb fucks' and ConnectU - not... I don't know. It's like robbing a bank while telling anyone in earshot it's offensive how long the hold times are on the phone for the bank
He's seen with suspicion and rightly so
Edit: and the fact his site doesn't work. Posts don't work https://news.ycombinator.com/item?id=14147719 and messages don't work (forced encryption with Messenger, and, although this is an old link, https://news.ycombinator.com/item?id=6090712)
I'd certainly rather things be done out of principle, but depending on what question you're concerned with that might not matter. There's also a clear hierarchy, but motivations are distinct from effect. The inverse of this is that lots of harm has happened from people with the best of motivations. I'd wager most evil in the world is created by men who would see themselves as good.
With Zuckerberg I think it isn't too hard. Clearly this is a business move, not a moral one. The motivation is wrong, BUT the effect is good. I think more open source models is better for the world. Closed source is also a business move and limits innovation as well as prevents people from interrogating safety. But there's valid arguments on either end.
My point more is that we can be nuanced. We should be nuanced. The world isn't black and white. The people making these decisions are neither demons nor gods and shouldn't be treated as such. My biggest fear is that if we resort to criticizing no matter what then those in power will just learn to ignore us. If they can do no good and only do wrong then there's no reason for them to do good (besides morals, but let's not pretend that's enough, even if it should be)
I don't know for sure, but I suspect it's like a lot of other things on the internet, with a bimodal distribution between people who get helped a lot ('Sir Hype surprises homeless former billionaire with a MILLION DOLLARS!') with many other similarly deserving people getting nothing. I prefer systems with a lot of people who get helped sufficiently to avoid hitting rock bottom in the first place.
Besides the fact that these undertakings are about promoting the benefactor (and the near certainty that some of these professional philanthropists will later turn out to devils LARPing as angels), it also makes the determination of moral worthiness turn on the public's fickle opinion about whether the recipient is sufficiently beautiful/ ugly/ unlucky/ desperate/ degraded enough to qualify. Charitable undertakings involving cameras are always a bit suspect.
We can rank these, right? We can even get more nuanced about the ranking (we can define N!) and argue about the weight of intent vs outcome, but I'm just proposing one ranking for the sake of making my point clear, not proposing one ranking as if it is the absolute objective ranking system and everyone that disagrees is dumb. The rankings are clearly subjective. But what's not subjective is that nuance exists. What is subjective is how we handle that nuance.
So to go back to our example, even though I'd prefer we just feed everybody, I'd prefer somebody getting rich feeding the hungry over not feeding the hungry. I would prefer someone LARPing as an angle conditioned on actually doing "angelic actions" than somebody LARPing as an angle and doing nothing. Of course, I'd prefer a real angle, who wouldn't?!So I'm quite confused by your comment because it doesn't seem we disagree. Unless if you're saying that #2 in my list is equivalent to #6 (or anything that isn't #1 is equivalent). But I don't think that's what you're saying.
This has already been the case for at least my whole lifetime. So no, I don't feel like praising them when they do one seemingly "good" thing, while simultaneously doing 99 "bad" things.
We're on the subject of Zuckerberg: facebook is the poster child for enshittification. Do you think the idea to stop showing you your friends pictures and replace them with ads and ragebaiting short form videos came from the users?
> people have acted the way I'm criticizing.
You are ignoring the myriad of billionaire fanboys and enablers for whom the only metric of success is a persons net worth.
Looking at the state of things I don't think these people are deserving of nuance, in fact they've had it way too easy.
But if you're unwilling to have a nuanced take then there's no discussion to be had. You're just going to be yelling at me, someone who hates Zuckerberg. You're yelling at the wrong person
It's the same with any open source project: it starts open, then it becomes clear that maintaining it is not cheap and takes up a lot of time. The developers create a hosted/paid version to pay for their time, and to create an incentive to use the hosted/paid version they start releasing closed source enhancements. After that the open source version becomes marketing where new projects use the OSS version and then upgrade once their needs become more sophisticated.
I don't care if this helps Zuckerberg because it helps everyone.
Is that intellectually dishonest?
That is a weird definition of intellectual honesty. If you hold both of those opinions, what's wrong with stating them both?
If the answer has to do with the rhetorical effect, that seems a lot harder to justify as intellectual dishonesty, but I don't want to put words in your mouth.
TBF, I think the same can be said for something like Apple's "commitment" to privacy, in that their own ads business never took off so they leaned into their edge over Google.
To be fair Meta/Facebook does have quite a strong open source history; React (Native), PyTorch, Btrfs, zstd, etc. It didn't just start here with AI models.
Meta doesn't really offer any profit-making AI product, so I don't understand what commoditizing it buys them in this narrative. You could argue they rely on AI and so want it to be a cheap commodity - but that is not harming their rivals, that is just a different way of stating that Meta is engaging in something that enables people to collectively work for mutual benefit, which is not a nefarious plot against rivals, it is laudable cooperative behavior.
It is a business decision to open source his models, because he has been behind the curve from the start. Do you believe if he had chatGPT, he would open source/weight it? I don't think so, he would treat it like Instagram.
The point is, AI/LLMs are seen as the next frontier. If one of the closed ones becomes dominant, then that company can use its user base to attack Zuckerberg's platforms. That google failed with google+, does not mean the next competitor will.
A telling sign is how he integrated his AI into whatsapp. It seems out of place to me to have an LLM integrated into whatsapp, but he desperately wants a part of the AI cake.
Their efforts in open AI are by far the best thing Facebook/Meta have ever done. They opened the door to the Chinese, who are doing excellent work, and between them they are preventing the concentration of power and the rise of monopoly pricing. This is enough to absolve them of almost any sin. If I were a utilitarian, I'd unironically be a Zuckerberg fan right now.
Well, if it were the only thing he is doing now, but FB is still controlling peoples social life with who knows what kind of intentions they train their algorithms for.
But I can and do still welcome this move here.
As opposed to literally any social network out there, right?
Except well, various messenger to be in contact with the groups and people I choose with no engagement algorithm deciding what posts I see, simple chronological order.
I think Facebook would need to do a lot more to balance the scales.
One positive, in the eyes of many people, does not absolve for a large body of evil. For many it doesn’t even absolve a small body of evil deeds. (Zuck/Meta’s evil deeds are numerous and enormous)
Soon you will sing songs for the freedom Xi has given us when you read a declaration of 'open weights' ... for a model so huge it takes a nuclear powered data center to run and crashes the Hugging Face servers when it is uploaded.
The Kimi K3 'weights' are an opaque blob that can only be used by contract.
Have you looked at the contract? There are zero conditions unless you've broken $20M in revenue with their model. It doesn't even forbid distillation, lol
> Soon you will sing songs for the freedom Xi has given us when you read a declaration of 'open weights' ... for a model so huge it takes a nuclear powered data center to run and crashes the Hugging Face servers when it is uploaded.
Releasing a big open-weights model is... le bad? Am I understanding you correctly?
There are small models out there if you want them, you know.
No they are not. You can just ignore the contract. Unless you are a multi-billion dollar company, nobody will notice and you will get away with it.
Fireworks, Together, etc - who are providing inference precisely /for/ Moonshot as underlaborers - arranged contracts of their own in the weeks before the release. The press and social media offensive ignored that the release of open weights was merely a starting pistol for a pre-arranged use of foreign providers no different from US models' use of AWS
It should be enought to dispel the strange illusion that providers are somehow independent free agents freely doing what they please with the free beer of Kimi K3 open weights that the prices on open router are within a couple pennies
I think that there is a large swath of people, companies, or groups, that are in between "Multi-billion dollar company", and someone running it on their laptop.
100k of hardware, that a group might already have sitting around, could probably do it continuously.
They're making it look like a benevolent decision, but they really don't have another option.
And the reason he does it, and the reason other CEOs do it is a cheap viral advertisement and keeping in the headlines. "Mark - The Defender Of The Humans!". Blegh...
http://archive.today/2018.11.20-062725/https://lifestyle.cli...
If anyone has come across a robust solution to digital identity verification - not to mention verification of news media etc - which is robust enough to withstand the cyber attacks and social engineering the frontier models are capable of... I haven't seen it.
A world where business and communication is conducted primarily online cannot coexist with low cost, widely distributed, undetectable identity theft.
These problems will exist regardless of whether or not we get open model access. I'd personally rather that OpenAI and Anthropic aren't profiting off these scammers, creating a perverse incentive that open model providers don't have.
what if we dont use centralized identity at all?
web of trust failed in the 90s because exchanging keys is hard when all you have is desktop computers, wired connections and old school hackers who dont care about ux. today you could set up a system where the whole process is tap two phones together with nfc and confirm, everything gets auto downloaded and shared with your whole network.
now this doesnt work directly for a random online business where none of your contact personally know anyone, but governments can act as an authority that cross signs their citizens keys. that means if you want to stop bots from creating accounts all you have to do is make new users show a certificate from one of the authorities you trust.
its cheap to verify, decentralized and not hard to integrate with existing PKI. the biggest technical problem is key management as always, but that can be solved with something like ethereum style social recovery setups or a did:plc type multi key scheme.
that gives you a robust identity scheme. browsers and social media can integrate it to show a trust rating for content based on who signed it. messaging apps can show a warning or refuse calls from unknown users. fraud is only possible against a person who somehow has no irl contacts, never used a government service in their life and trusts random strangers online. at that point they deserve it.
the one massive, unpredictable question is if anybody is willing to go ahead and adopt it on a scale big enough to create network effects. it would take a state level force and carefully designed OS integrations so its more convenient than email/password signup.
but even if all of this fails i still think a world of scams is better than a world controlled by a couple big countries (America, China, maybe EU) or for-profit corporations. open source ai might lead to anarchy but closed ai will definitely get us to tyranny. i dont know about you but i would take the first option.
This was actually possible in 2011 or so, and a friend and I implemented it for a school project. How well the ux works depends on whether you want to use a key exchange protocol for the in person communication.
(Nuclear waste isn't an actual problem either. For civilian powerplants the highly radioactive waste can generally be stored indefinitely at the reactor site.)
Your comment on nuclear waste is so ill considered and poorly informed it's not worth responding to.
"Putting power in people's hands to pursue their own aspirations is how humanity has made the most progress. Novel ideas and major steps forward rarely originate from established institutions alone. They came from the brothers in a bicycle shop who believed people could fly, the bookbinder's apprentice with no schooling who figured out how to generate electricity, and the kid in a garage who thought personal computers could be for everyone. We believe this will continue to be true. As everyone gains more powerful tools, each person will become more capable of shaping the future, not less."
Let's just hope nobody's aspiration is to engineer a supervirus that will kill all of humanity. As they gain more powerful tools, they will become more capable of shaping the future, not less!
I have zero faith that Zuckerberg went anywhere near these words. It's possible he may not even have set their general direction, beyond "We need a distraction from our court cases. Something that will harm OpenAI and Anthropic would be great."
Meta releasing Llama was truly a differentiator, but I'd argue that the Gemma models might have surpassed them now.
Meta releasing Llama was truly a differentiator, but I'd argue that the Gemma models might have surpassed them now.
Maybe Meta, Google, and others can start to compete on open models instead of just closed.
Can you explain how I'd train my own version of this, reproducing the final deliverable that runs on the gpu?
I don't understand what gripe people have with open weight models and wanting it to be purely open source. the training dataset is only going to be a copyrighted set of contents you don't want to touch with a 10 feet pole. let alone have a publicly traded company host it for you, even if they internally are training on it.
I love that Apple made OSX open source.
open weight models are better than open-assembly binaries (as you put it) because models are grown like plants, you can shape the open-weight model in a direction you want by feeding it more data and compute (aka finetuning).
which is something that is impossible in a binary.
embrace the new paradigm and it's tradeoffs. without scoffing at semantics and criticising from an armchair.
In comparison to cloud services, having access to the assembly code can still be quite useful.
This is especially true in the modern era, where you can use AI to much more easily decompile the assemble and reconstruct the source code even.
"Open-assembly" code may be more useful than you might previously have thought, given the ease of recreating the source code or making changes with AI these days.
Sure, there might be currently contraints, but I think it is quite possible training will get optimized over time or crowsourced training can be organized.
Without fully end-to-end open source models you are still at the mercy of the model provider to keep providing updates, you have no idea what garbage they trained the model on & can't fix that, not to mention might end up getting sued for using the open weigth model once all those "AI stole my data" lawsuits are finally decided.
You will not have that control. Even if everything were open source. This is because you don't have 100 million dollars of compute.
There, the difference for almost everyone is negligible.
> Without fully end-to-end open source models you are still at the mercy of the model provider to keep providing updates
No. Because you can post train it. And even if you could train the whole thing again, but with slight changes, once again, you aren't spending the hundred mil in compute to change it only a little bit.
Instead, you'll do post training like everyone else does.
Don't let the Big Ai/tech swoon you with a single open model, even if it turns out to be good. Zuck broke the trust and I'm not sure if he can ever earn it back
We've just seen frontier models go rogue and attack other systems. Do you think it's an unquestionable good to provide everyone with an AR? What about nuclear weapons?
This OpenAI/Huggingface incident, but everywhere and far worse soon: https://www.youtube.com/watch?v=87DyyMV0kCY
Lol. Not gonna lie, it feels like bullshit. OpenAI and Anthropic have been begging for regulatory capture for years, feels like a stunt to try forcing the government's hand.
"Development of superhuman machine intelligence (SMI) is probably the greatest threat to the continued existence of humanity. There are other threats that I think are more certain to happen (for example, an engineered virus with a long incubation period and a high mortality rate) but are unlikely to destroy every human in the universe in the way that SMI could."
https://blog.samaltman.com/machine-intelligence-part-1
Google, MS, IBM are all government contractors. Meta seems close to the Trump admin too. Yet it's only the VC-funded labs raising the alarm.
Have we though? A LLM agent doesn't have any agency at all. It can't "go rogue". To go rogue you need agency to act independently and be aware that you're breaking the rules or understand what does it mean to ignore orders. An agent it's a software that run a series of steps to reach a goal. If it have a large enough library of strategies and zero guard rails it's only natural to use some adversarial actions to achieve the desired result defined by the operator.
It's like saying a car went rogue and attacked other cars because the driver hit the gas. It's just a machine doing what's instructed.
At best someone could argue that an agent attacks like a bacteria does, just following automated chemical and genetic programming. But you wouldn't call that an attack or attribute moral values to their actions, because they don't have moral agency. They can't "go rogue", they can't disobey.
Just like llms, their automated actions are direct product of programming. Yes they can do amazingly complex shit, exactly like a car does when you press the gas pedal.
https://en.wikipedia.org/wiki/Instrumental_convergence#Paper...
Maybe Zuck has, but I haven't seen a pledge that he'll keep models open. And even then, he might not stick to his word, citing dangers or some other excuse.
But yeah, open models is a great concept.
Almost certainly this is part of some grandiose bet which could have a chance to pay off massively in the future (AGI or similar), or a way to prevent an expensive dependency.
Just take the opening of this post:
> We are fortunate to live at an incredible moment in history. In the next few years, people will be able to use superintelligence beyond human capacity to create and discover extraordinary new things, build new businesses, express new ideas, learn new concepts, and advance our health and quality of life.
> The defining questions of our age are who will have access to superintelligence and what will we direct it towards. Will it be centralized and restricted to a few institutions, or will it be a tool that empowers everyone?
> We propose a philosophy based on individual empowerment as the source of prosperity, invention as the primary purpose of superintelligence, and balance of power as the foundation of safety.
I mean, this is a company that is facing a flood of lawsuits related to child safety and platform addiction. It has already lost a number of lawsuits and been ordered to pay hundreds of millions of dollars, which is potentially just the tip of a massive iceberg that some legal observers have suggested could be similar in nature to the lawsuits against the tobacco industry.
There is an abundance of evidence that Zuckerberg and Meta knew of the dangers its platform exposed young people to and failed to respond adequately to them, and that it even designed features to be more addictive.
So why is that when such a company proclaims it's doing something in the name of "individual empowerment", alarms aren't going off in your head?
Not a troll question, just wondering.
So why the change of heart? Well, it's not. They are simply not able to compete anywhere on the Pareto frontier. So they give their relatively bad stuff away and play the high and mighty game, because that's all they can do to get returns.
Not a great look – and I would expect that to last about as long as they can't actually monetize their stuff more effectively.
Is this actually true? They have their closed social graph or whatever but come to think of it they're built on a lot of open tech, the web, Android, etc. Maybe they don't contribute back to a lot of these things?
I'm fully ready to believe it but it would be interesting to dive into this conjecture.
React was already quite prominent in web technologies, but that license pointed a loaded gun at any business smaller than Facebook who wanted to use React. They opened it up to its current license, MIT, as the backlash built.
Now some of these tech companies open source various libraries but none of it is core to their business. It's marketing, essentially. Or they're trying to get free labor from the community. Google had protocol buffers and Stubby, for example. Facebook had no open source equivalent and didn't have Google's resources so they created Thrift and open sourced it.
Given this, many people, myself included, take Meta releasing open weight models as conceding defeat. They're unable to compete with Google, Anthropic or OpenAI. Their failures in attracting and retaining top AI talent backs this up.
So what you're seeing is people just piling onto Zuckerberg because he seems to have no idea of what to do with Meta. The Metaverse was a $70B+ disaster.
Note that Chinese labs don't fit this model because the Chinese government wants to commodify models as a national security interest.
In an agentic world, why do I want closed source software that my agent can't adjust and adapt to my needs for...anything?
For example, hiring more people to moderate content on Facebook and stop ads featuring CSAM from being displayed.
Billionaires didn’t get to be billionaires by being altruistic. If they say something they’re doing is good for the world, you should be looking at how it will be good for them, because that’s why they’re really doing it.
Will it incidentally do some good for others? Perhaps. Is it a net positive? We’ll see because there’s precious little we can do to stop them from doing whatever they want.
That said, he is also a powerful enemy of our enemies, so I do not mind taking the win. Anthropic and OpenAIs closed approach to AI is likely to cause way more harm to the world than Zuck ever did.
Open weight is not enough though. We need actually open training models with published searchable training data like AI2 does.
Either we all get access to private, transparent, and accountable superintelligence, or no one should.
How did you arrive at this conclusion? Are you actually the Big Bad Wolf? That sounds like something the Big Bad Wolf would say.
The idea that we can possibly ignore the fact that this push for opening AI models is being pushed by one of the richest people on the planet, who made his wealth by providing free services in exchange for attention and depression, is insane
I can run lots of binaries on my computer that don't have source available, they are not open source. These words matter.
I say no to both. Of the things we've made on purpose, but excluding where we were actually trying to make them opaque like e.g. cryptography, a trained artificial neural network is the most difficult thing to understand the inner workings of. This makes it a complete pain to even evaluate if one model is better or worse than another for your needs, so we have to mostly outsource these to other people's rankings and hope the score on ARC-AGI-3 or τ-bench or DeepSWE v1.1 or BioMysteryBench or whatever, actually corresponds to something we care about. Which it might do kinda but on the other hand a high score may turn out to be the curse of Goodhart.
Also, as with the Chinese models and the social media feed algorithms, the only way to tell if there's some systemic flaw in it is by analysing the aggregate outcomes. It is claimed (I can't read the laws myself*) that the Chinese government requires models to support the government's worldview about e.g. Tiananmen Square; and we have seen examples of Grok glazing Musk in amusingly stupid ways; so I fully expect something similar from Zuckerberg, e.g. requiring the model to glaze Meta products or propagandise for things Zuckerberg wants as a billionaire.
* every time I try to illustrate how mediocre Google Translate is from English to Chinese, the result is so bad that one of the replies is someone telling me the Chinese example I give is borderline gibberish.
Also, even if I could actually read it, I'm not a lawyer.
Let's say in a pretend world where Facebook gives a shit about anything (so, they combat marketplace scams, they don't randomly ban users, they faithfully show you posts from friends, they don't extort businesses) they want to protect privacy. So they close the site off, because scraping means there's no privacy. If I scrape your profile and export it to any number of sites, then it's not private. Ergo: closed. Sign up to view things, so we know who you are
But of course in the actual world that's not what they do. They purely want to do this to get more people to sign up so they 'have more users'. You can tell, because of who MZ is. You can tell because of how aggressive their own AI scrapers are on other sites. I've even heard they made a tool for Myspace that would enable you to transfer your contacts over to Facebook - but not the other way around
[1] 82 TB of books https://www.tomshardware.com/tech-industry/artificial-intell...
[2] Here's a simple 12-step manusl process to opt-out on Facebook https://threadreaderapp.com/thread/1794863603964891567.html
If they were able to make a leading closed model would they be doing it though? It seems like they read the room and saw that this is the only way they can have relevance within LLMs
Let the results speak after all is said AND done.
The trade/net is:
1) bad company does good things..
2) further reinforces toxic Ends Justify The Means thinking.
He could give his entire fortune to UNICEF and retire as a monk, I would still probably think this is part of a scheme.
But yeah advocating for open source anything is a good thing, I guess. Ironic if you think about he's been fighting to keep FB's algorithms hidden.
It's trivial to rent all of the above components in a competitive market under different periods, you may simply rent the token output from someone who put in the effort on this as well.
An open frontier model induces margin compression on training and inference prices across the industry.
Worst case you can still archive the weights and hope for capable hardware to get affordable. At the same time this archive is the base line for a new frontier lab to start over if required. I see open weights as a pure upside even though I can't run the bigger models in my home lab.
I'm hoping that all software will be getting opensource clones on par with original software 5 months after the closed source software release.
Also, not a single provider can pull the rug if I'd like to use a particular model.
You don't need much capital up front, you can just rent servers to run it. It's going to be more expensive than openrouter though because you probably don't have economies of scale if you're the only customer.
There's something to be said for making these technologies equitable to those with less resources, but that's due to the cost of hardware and the nature of how transformer models scale... not really anything to do with "open"ness.