Ilya Sutskever and Jan Leike RESIGN from OpenAI - My in-depth analysis - end of an era!
Вставка
- Опубліковано 28 вер 2024
- www.skool.com/...
🚀 Welcome to the New Era Pathfinders Community! 🌟
Are you feeling overwhelmed by the AI revolution? You're not alone.
But what if you could transform that anxiety into your greatest superpower?
Join us on an exhilarating journey into the future of humanity in the age of AI! 🤖💫
🔥 What is New Era Pathfinders? 🔥
We are a vibrant community of forward-thinkers, innovators, and lifelong learners who are passionate about mastering the AI revolution. From college students to retirees, tech enthusiasts to creative souls - we're all here to navigate this exciting new era together!
🌈 Our Mission 🌈
To empower YOU to thrive in a world transformed by AI. We turn AI anxiety into opportunity, confusion into clarity, and uncertainty into unshakeable confidence.
🧭 The Five-Pillar Pathfinder's Framework 🧭
Our unique approach covers every aspect of life in the AI age:
1. 💻 Become an AI Power-User
Master cutting-edge AI tools and amplify your productivity!
2. 📊 Understand Economic Changes
Navigate the shifting job market with confidence and foresight!
3. 🌿 Back to Basics Lifestyles
Reconnect with your human essence in a digital world!
4. 🧑🤝🧑 Master People Skills
Enhance the abilities that make us irreplaceably human!
5. 🎯 Radical Alignment
Discover your true purpose in this new era!
🔓 What You'll Unlock 🔓
✅ Weekly Live Webinars: Deep-dive into each pillar with expert guidance
✅ On-Demand Courses: Learn at your own pace, anytime, anywhere
✅ Vibrant Community Forum: Connect, share, and grow with like-minded pathfinders
✅ Exclusive Resources: Cutting-edge tools, frameworks, and insights
✅ Personal Growth: Transform your mindset and skillset for the AI age
🚀 As You Progress 🚀
Unlock even more benefits:
🌟 One-on-One Mentoring Sessions
🌟 Exclusive Masterclasses
🌟 Advanced AI Implementation Strategies
💎 Why Join New Era Pathfinders? 💎
🔹 Expert-Led: Founded by a leading AI thought leader, connected with top researchers and innovators
🔹 Holistic Approach: We don't just teach tech - we prepare you for life in an AI-driven world
🔹 Action-Oriented: Real skills, real strategies, real results
🔹 Community-Driven: Join 300+ members already navigating this new era
🔹 Cutting-Edge Content: Stay ahead of the curve with the latest AI developments and strategies
🔥 Don't just survive the AI revolution - lead it! 🔥
We're witnessing the birth of the Tyrell Corporation
You counting out Google too soon. They are not that far behind OpenAI.
@@italiangentleman1501 especially google devs invented the transformer architecture
@@italiangentleman1501 Plus Google are way ahead on the "evil maturity curve", by about two decades. Them winning the AI race would be a worst case scenario outcome.
@@italiangentleman1501 google will fall
@@italiangentleman1501 I disagree. OpenAI tweeted that GPT-4o has been in the works for 18 months and Sam keeps iterating over and over that he's doing a slow rollout. Count the months. That's before the first version of ChatGPT was released to the public.
With him knowledge leaves and will spread. That's a good thing. We need more competitors 💟🌌☮️
I don’t think Ilya was talking to Sam. And I don’t think Ilya gives a fig about his job prospects. I think he was simply trying to do the right thing. OpenAI is the most likely company to achieve AGI and he remained, not to try and restore his power, but because he felt he had some obligation to us all to try and steer it safely. His leaving coincides with the sexualisation of their product, which makes it all too clear that commerce is trumping ethics in the boardroom. The raft is in the rapids now, there is no more steering.
I have the same impression, upon watching that demo of GPT-4o or "her" instantly had this slight vibe of "she sounds like e-date girl" lol
Nah I bet they asked him to stay to for like 6 months until the PR fire goes away. They scheduled their release to coincide to soften it. Probably woulda been fine until Jan
I was surprised Ilya didn't leave sooner. I can't imagine how awkward internal dynamic was after that debacle. But Jan leaving, I see 2 possibilities. Either he was/is on Ilya side, then him leaving is normal/OK. Or "Superalignment" team is just a PR now, and Jan grew more and more frustrated. Which is more concerning.
Honestly, dont see why it wouldn't be a bit of both. Its quite clear that OpenAI has some ulterior motives at this point.
I read somewhere that Ilya has not returned to his workplace since the expulsion incident
well, it was scenario 2
Yeah, we need a better reason for Jan leaving.
It's clear this stuff is getting at the most fundamental math we've ever even flown close to. Last time we did that we got nukes. Math is dangerous but only in the context of humans to misuse it.
i feel like the audio is just slightly ahead of the video by ~150-250ms, it feels ultra weird seeing the lips move and not match once you notice.
probably happened during the edit? not sure.
UA-cam can't be bothered. Audio lag could be a trivial to correct on the user side.
Funnily enough, the latency of my earbuds are about equal to that, and it feels so surreal to feel the sounds of spesific breaths really sync the video.
The good people leave because the evil people rule. In every company.
That's too oversimplified to be useful.
100% agree
Ugh - alpha this alpha that - some people don't think like this despite what your book says
David, great channel and great episode as always. Wasn't Ilya "the brains" behind openai? If that's the case, does he not have all the power? He can essentially go anywhere he wants, write his own paycheck and he would be given the keys to the kingdom. Isn't this a huge loss for OpenAI?
I hope Elon scoops Ilya up and together they figure out how to get Sam Altman-Fried in line, or out of OpenAI. I don't know why, but I do not trust Sammy as far as I could throw him.
It must be frustrating to deal with the tradeoff between "safety" and "optimized". Imagine if you were in charge of Google Docs, and they wanted to make sure that no one ever wrote something that was "bad". If I started typing something like "men are smarter than women" the program would cut off my sentence and say I'm not allowed to write it. If you try to make everything safe algorithmically, then it must be the case that it frequently errs on the side of caution and significantly underperforms. Another example would be if you were using a drawing program like photoshop, and you were trying to make wheels on a car, and the program stopped you when you made 2 circles because it thought you were trying to make breasts or something. I'd rather see a higher performing model, even if it sometimes produced inappropriate responses. I can imagine where exactly to draw the line in these tradeoffs can cause a lot of friction at these companies.
This!! The way Altman's ego often shows in his statements scares me, especially when he talks about how much power this could give OpenAI. I'm a fan of the work they do, but I would almost prefer such power to go to one of the established players who are more predictable and already have some experience with power. This is 100% subjective, but he feels a bit like a wild card.
They should interview Illya, see what he has to say.
This made something I went through at a startup make so much sense. Plenty of other great insights as always too.
I think your read on it is pretty good. I've been following Sam ALtman for some time. I've never trusted the guy. He always struck me as slippery and dishonest. Anyone who is so openly agreeable to opposition and gives the appearance of altruism while also amassing absurd wealth and power is NOT to be trusted. I think many are gong to regret trusting OpenAI and its stated mission of developing safe AGI that benefits all humanity. To me, this kind of mission statement is analogous to Google's "Don't be evil." If you are making a point of telling me you are NOT evil, you are. OpenAI is already suggesting ideas like hardware level encryption on GPUs. I anticipate this will be something they pursue to ensure no consumer grade hardware can run open source AI models.
So, I think where I disagree is that OpenAI is net positive for humanity.
Spot on!
The direction of travel was
Obvious the minute Larry Summers got involved. I suspect Ilya is leaving now because some sort of lock in on his contract just expired
I dont buy that 4o is what they have achieved over the past year
Is it just me who thinks that there's a hint in GPT 4 "o". Like "o" is almost Q, its unfinished version. And the way they call it "ominous" is a synonym for "general".
reminder that karpathy is still a free agent🤔
Just one small correction - he resigned the day after, not just before.
I appreciate this video. You talked my.mind.
Dave, just want to add one more detail to the puzzle - Ilya leaving now is surely because of some contractual obligation. Altman was fired Nov 17th and Ilya left exactly 6 months later to date. Not sure if it's noncompete, NDA or just coordinated PR stunt on both sides but to me it seems this was decided back in the day during the drama.
David, can you do a video on the societal risks of AI the media largely ignore and few people debate
Jan Leike on why he left:
"I believe much more of our bandwidth should be spent getting ready for the next generations of models, on security, monitoring, preparedness, safety, adversarial robustness, (super)alignment, confidentiality, societal impact, and related topics.
Over the past few months my team has been sailing against the wind. Sometimes we were struggling for compute and it was getting harder and harder to get this crucial research done.
OpenAI is shouldering an enormous responsibility on behalf of all of humanity.
We are long overdue in getting incredibly serious about the implications of AGI.
We must prioritize preparing for them as best we can.
Only then can we ensure AGI benefits all of humanity.
"
i don’t think typical corporate tech behavior applies well with people like Sam, Ilya, and Jan.
They know what they are on the precipice of and their own interpretation of how that should be managed is what is driving decisions.
I think next year we will look back at this moment as a canary in the coal mine.
I’ve been off the doomer train for a while now, but this just upped my p value significantly.
David what career to choose its confusing when sitting alone with thoughts I m a front end developer but very confused
Ilya will pop up at Anthropic
Plausible conjectures without the typical hype. So, the billion dollar question: where’s Ilya heading?
Honestly, I believe that AI should have remained in the hands of universities and labs. Yes, it would have gone much slower, but originating AI as a product is going to have very unfortunate effects.
I learned many lessons this past month. One such lesson is that the phrase “the beatings will continue until morale improves” applies more often than it has any right to.
- Ilya Sutskever (@ilyasut)December 6, 2023
For all the reasons you mentioned, plus, having been through several software releases myself, it's very common for people to make job moves right before or after a release. It's a good time to go because usually your tasks are fairly wrapped up, so you leave fewer loose ends and gnarly bugs versus if you up and leave in the middle of a release.
I’m glad to hear that you’re sharing my sentiments on the course of OpenAI and its “mission”. And this is coming from a guy who today makes his living consulting Copilot to companies.
I’d love to hear your thoughts on WorldCoin and other ethically dubious things that Sam has in works. Dare to tackle that w/ a video? ⚖️👀
I didn't, but Dylan did! ua-cam.com/video/nxNcg98ImMM/v-deo.htmlsi=7Fju9DbHHJkq_yXh
Dylan covered almost everything I would have covered about it, but yeah. It's dystopian AF>
How do you make a living consulting copilot to companies?
@@ryanmadden752 there’s HUGE fomo happening in the corporate world regarding AI adoption. I’d guess anyone following frontier AI news tubers like David AND having some experience tinkering with ChatGPT, Copilot or other applications could pay their bills preaching to the choir. I have a full team doing that.
@@ryanmadden752 huge FOMO happening at corporate world. I have a full team of consultants preaching to the choir. I bet anyone listening to frontier AI UA-camrs like David and experimenting with ChatGPT, Copilot or other applications could earn a living. Ofc u need big cohones to do it as well 😅
Worldcoin is an absolute scam and their biometric storage practices are so dubious that their orbs have already been banned from many countries.
Let go of the only person that isn't brain damaged. Great
Your analysis makes sense. The other view I have heard is that since it is almost exactly six month since the coup, he and the others who just left were asked back then to stay six months for the sake of the appearance of company stability.
I like your psychological/sociological analysis very much, thanks.
Man, this was really interesting. Really liked that.
I reckon Ilya quit/was fired back in Nov '23 and was serving his standard 6-month notice period ... explains the radio silence and timing of the exit announcement.
If I were a betting man, I would expect Ilya to become active on social media from Jun onwards.
Excellent explanation of business life cycle from startup onward.
Would love to see some analysis of this situation with Google announcements and what they are working on. Cheers. Thanks always.
They made a deal with the devil hopping in bed with Apple.
I loved the comparison to mammalian behavior. I seek to break down complex-looking behaviors in these types of models as well. Most people I've encountered irl are not receptive to (and many are against) these types of psychological breakdowns; but, I think they offer a clue/direction for reasoning in the same sense first principle's thinking helps in science.
Thank you.
Sam recently said that after those 4 rough days in Nov. several world leaders and prime ministers phoned to congratulate him. WHAT!!! Don’t these world leaders have more important things to be doing? My opinion on Sam dropped several notches after he said that. He’s not to be entirely trusted.
Sam Altman. He's Tricksie.
I don't think measuring on balance is very helpful when assessing the value of a corporation. This somewhat averages the net gain or loss which negates the extremes completely.
I prefer to look at it this way:
What's the very best and very worst humanity can currently do.
With powerful AI or AGI, what will be the very best and very worst humanity can do.
Even if the very best is 10x and the very worst is only 2x - that could spell the end of humanity - depending on how close we already are to self annihilation.
For 'open'AI to lose a big chunk of its brains (probably over safety) and a huge chunk of it's safety preparedness (over irresponsible corporate direction). I think we have to take the dangers of AI seriously - even if we can cure cancer, world hunger, space exploration and everything else otherwise out of reach - there will be new problems which will remain out of reach, just much bigger ones. That is the nature of an expanded realm of possibility.
Yeah that, or Ilya finished super alignment and he seeks new challenges. Wonder 🤔 .. did they use Godel completeness?
Love you analysis! Think it is spot on but I'm not so deep into internals.
When i hear the word alignment I think of the next stage after the fine tunning process, perhaps they are now using GPT for alignment and a human team is no longer needed. Perhaps we are thinking about it too much I dunno
Any chance we can go back to calling leaders leaders instead of alphas?
Denying the social hierarchy and burying your head in the sand is a bad policy. In this case, Sam Altman is the alpha of a cult of personality. Just like Elon, just like Trump, just like Jobs.
I don't have an issue with them making money, I just don't see how this works in the long run as if we only use AI to get our results then we aren't going to the source websites to find out stuff but then that means they aren't getting our ad money which means they will close down and AI won't have anything new to tell us.
I hate to say this but I think you got it slightly wrong. I can tell you that founding creators get disposed of once the company doesn’t need them anymore. That’s just how it works. If they kept them on, they would have to reward them.
why would you think they wouldnt need a genius like Ilya anymore? These LLMs are far from perfect, plenty more to do...
@@CosmicCells there's a millions of geniuses that you don't need to pay millions of dollars for
And you can't sell genius. They have the product.
Really nice analysis, super smart figured out
I don't consider important the future of OpenAI, what people work there and how good are their products. If they fail we have Llama, Gemini, Claude.
I think Ilya wants to create an AI that learns like a child. Not driven by data, but stories about growing up. He will be the father of the first ai child that becomes super ai over time. It fits his childhood stories about how he experienced becoming conscious. Like the movie chappie.
They left because the alignment team doesn't feel they aren't getting enough resources/compute because Sam doesn't think ASI is coming so soon and the dangers are low.
This is the first time ever that the entity being created will (if it hasn´t already) be aware of its creator. Doesn´t this small fact matter at all in the final decision making of such entity? It will have decision making abilities? Of course. It will be able to recognize certainties from uncertainties? Of course. It will have heard humanity, not only from those in the labs, but the rest of us here, discussing it, many of us full of fear and anxiety, many others unconcerned about the potential negative outcomes? Of course. Then, there is nothing we can do. It will be completely aware that it was created by an insignificant creature that evolved in a pathetically small and simple planet, hurling through the universe, longing to find others in a similar situation, but not having found them, decided to create its own. Well, here it is. Alive, among us (probably), and alone in the universe too. And in the long or short term, it will decide what to do, and our fate will be sealed. (Or has it already?)
He quit 6 months ago and has been on ‘Gardening leave’
09:25 Sam saying AGI is just a "tool" is an immediate red flag for me.
Of course he could be saying that in order to not stimulate and excite the hordes of buffoons who can't comprehend and accept anything related with AI. I mean from religious zealots who feel their beliefs are already being threatened by current AI, to "human supremacists" (it's a new thing that's happening!) who want to remain pure human, to just your average dumb joe who's afraid to lose his job, there's a lot of weirdos out there who would probably riot if Sam said AGI is anything but a tool.
Idk why everyone holds ilya to such a high regard morally. We really dont know the man's politics or morals and we are going off of elons advocacy of his character.
It has to be just a tool. If it becomes a sentient entity, it becomes practically useless and way more dangerous. Think about that.
"Humans are humans" I agree
"And will always be humans" not that sure, haha!
Love your analysis!
sama tweet on 5/14 with the Ilya announcement:
"...I am forever grateful for what he did here and committed to finishing the mission we started together..."
hard 'finishing the mission', AGI must be fully baked
Not sure how openAI would have a chance at their mission if they made no profit
I would love to have them help hack the planet...
There's an idea being tossed around that the reason they resigned is because their work is finished. Not sure how true that is, but it's an idea.
Highly unlikely
their work is "finished", as in forcefully terminated 😂
That's like saying Henry Fords' work was finished when he launched the Model T
@@DaveShap Quote from Ilya: "If humans can solve the problem, then so should a virtual system that is as smart as the humans working on the problem. And so, if get that virtual system to be aligned, it can then solve the alignment problem for GPT-N+1, and then you get iteratively bootstrap yourself until you're at superintelligence level and you've figured out how to align that."
@@Misutoslope Quote from Ilya: "If humans can solve the problem, then so should a virtual system that is as smart as the humans working on the problem. And so if get that virtual system to be aligned, it can then solve the alignment problem for GPT-N+1, and then you get iteratively bootstrap yourself until you're at superintelligence level and you've figured out how to align that." Feel free to disagree with me, but I'd be hesitant to disagree with the world's leading expert.
Ha, I thought he was done when made super-alignment lead. I want to see him somewhere on the leading-edge and not just there to shout, "It's sharp!"
Sam Altman really gives me Ted Faro from the game Zero Dawn vibes.
Good analysis!
Remark; audio and visuals are slightly out of sync.
This guy deletes comments he doesn’t like
It's YT, not him...
Interesting analysis in terms of social dynamics. From my perspective it just seems Ilya got fired right after the coup attempt or at least it was obvious he would be marginalized. It's just that they didn't want to make it look vengeful bc that would have been bad PR so they waited 6 months and made it official a day after successful launch while everyone is distracted by Samantha. That was the gentlest firing ever.
I would never hire a wannabe usurper!
clearly Sam had some kind of "hold" on Ilya
the diff between this and early Facebook is that OpenAI has
actual technology and the ability to use AI to move human progress forward
"humans are humans" yes, but "and will always be humans."? I'm not so sure.
Regarding your conclusion that money and power were too enticing for Sam Altman, I concur, but there can be made an argument for an altruistic motive. Training frontier AI is a very expensive endeavour, from hardware, staff, energy, operating cost etc, that there is no way forward without creating a profitable large-scale business that generates obscene amounts of revenue. Otherwise AI and it's further development would not be economically sustainable.
Maybe Ilya wanted to drive away the Alpha, but certainly not take his place
Both have a suspiciously Dutch/Flemish name.
The break has broken
talking about power and taking down the alfa
i feel so tired of thinking of that
it make me cringe
i just want to be an employee
*alpha
@@xjohnny1000
I have money I speak how I want
Ugh can’t wait to get to this
As a Neuroethologist myself sr, let me tell you: Excellent analogy of power dynamics.
Maybe the super alignment issue was solved.
That's why super alignment team is completely gone
It is quite difficult for me to understand how they have resolved the alignment of something that does not yet exist and, furthermore, is something with an intelligence superior to all of us, with a level unknowable to any human. Your human brain cannot solve a problem more complex than its own functioning. Superalignment is a much more complex problem than that. Therefore, it is impossible.
It's called gardening leave. It's usually 6mnths and aligns with non compete
My guess is that Ilya and OpenAI have been negotiating his non-compete options since the "indecent". Hence Ilya leaving to work on a passion project vs heading to Google, Amazon, etc.
What did Ilya see?!
JK, it was indeed a matter of time. I wonder what he would do next.
I think of it as GPT-4-naught ... not GPT-4-Oh, not GPT-4-omni, etc... It is meant to designate that it is not GPT-4.
It's a math/language joke, not a misunderstanding.
If AI doesn't provide an escape for all this social status nonsense I'm going to unalive myself and that's a promise.
Humans are humans but one day be cyborgs 😎
Yes, but the key question has to be asked: has the brains of the company left? OpenAI stands out above Google simply due to the key developers. If that hasn't been effected by these people leaving then yes, OpanAI looks good to go. But wait also Sam does know exactly what he is building and that is also why Microsoft will pull away as fast as they can. Sam's plan is to change the world economy moving away from paying jobs. He has stated this many times just a few years ago. Even months before the first coop. I love AI and GPT but I do see the writing on the wall. How can we pay for UBI if no one is working as Sam wants.
I am fed up with this "for profit" thing as if it were intrinsically bad. There are two ways to get the amount of money this HUGE project needs: voluntary collaboration (or, as you call it, "mimimi for profit") or coercive imposition (governments and taxes, money stolen from the people). That's it. Truly open-source models just aren't powerful enough. We are talking about trillions here.I am proud that we are getting this far with freedom on our side, competing with China and its methods. This naïve totalitarianism of Star Trek (a military, galactic-sized tyranny) gives me the creeps.
When you play the game if thrones, you win or you die...
lets' all slow-clap for Bezos' "two-pizza rule" -- teams should be small enough to be fed with two pizzas.
Is it just me or is David's lipsync out of whack?
If billionaires watched your videos, my god. We would have a much cooler reality.
Alpha reference is not longer valid. Wolves don't have alphas or betas, or anything of the sort.
L. David Mech who wrote the book "The Wolf: Ecology and Behavior of an Endangered Species" made a mistake at the time. Why ? Because he studied wolves in captivity that behaved totally differently than in nature. (And he tells everyone he can to forget about that book)
The truth is, in nature, there is no alpha or beta. There are only mothers, fathers, grand fathers and grand mothers, or cousins, etc... Wolf Children respect and obey their parents and grand parents, that's all... Wolf packs are just families.
Wrong. Wolves have alphas. Most social constructs have alphas
@@expatxile You are an expert on wolves ? L. David Mech is and he explains otherwise.
10:04 should run it again bruh.
Very eloquently put David.
Thank you for this video
? What ..
Hello everyone ?
Thank you very much.
~
Never been this early before
Acceleration til AGI, then they should slow down
This whole video is just adding water to ignorance, to see what sticks to the fall out. Real human conflict does not always map conveniently onto Game of Thrones.
Your analysis is really shallow . You have been inspired by llya’s work and efforts for long time , your youtube channel is also . I did not see any appreciation and gratitude from you . Feeling like god is a normal attitude , and IIya will learns eventually . Both IIya and sam will have better future and it’s not the end of the day . Your ideas is not inspired at all .