Sam Altman & OpenAI | 2023 Hawking Fellow | Cambridge Union

[@Andylium]

This is going so fast, that it’s been 2 days since the release of this video and Sam Altman’s no longer CEO of OpenAi.

[@johnkintner]

next week it'll be "no human can communicate with the board as well as ChatGPT can, so GPT-5 will be our acting CEO"

[@jerroldneal]

It's horribly ominous this happened. I didn't even think of this scenario: a thoughtful and careful CEO gets suddenly fired. That could be the beginning of the story where AI goes haywire.

[@jamessteele7010]

@@jerroldneal and the president quit right after. smh

[@bgill7475]

Microsoft will fully take over now.

[@Darhan62]

Sam will land on his feet. He's brilliant, well-meaning, and a good communicator. OpenAI still has Mira Murati and Ilya Sutskever, and I'm sure it will continue to focus on its mission.

[@Minimalrevolt-m83]

Thank you very much Sam Altman and OpenAI!

[@BenRovello]

I’m opportuned to have guidance and supervision from some of Wall Street’s eyes and ears. This is an exciting time of year after a tough one all round. I made a significant profit with my MSFT holdings recently and have drawn almost 100k for the holidays.

[@simone_maya]

That's awesome! How did that happen?

[@BenRovello]

Well, a renowned figure in Manhattan I had contacted by Monica Mary Strigle recently added MSFT to the portfolio, and things took off few days back. Plus, there's some interesting speculation.

[@xavier_lucas]

Speculation? Do tell!

[@BenRovello]

You won't believe it. Sam Altman, the guy ousted from OpenAI, got a warm welcome from MSFT. I saw it as a potential game-changer, and it seems I was right.

[@FranciszekPawal]

Sam Altman joining MSFT does sound like a big move. How did it impact your MSFT holdings?

[@tristanwegner]

Such a pleasure to have high quality question AND sensible answers to them

[@Techtalk2030]

This is going to be one of the most important decades in human history. What a time to be alive.

[@Bezimienny1598]

Remember that in the Matrix universe, the machines were incredibly peaceful. They only became aggressive after humanity attacked their city because the humans were scared of their technological superiority. If the world leaders in that universe weren't total braindead fucks, the humans and machines colud have peacefully coexisted.@@jeremy-b

[@benderthefourth3445]

Said every person in every decade...

[@AlexanderIsDead]

Maybe true if you're employed or have a business or some sort of income. Terrifying if you're unemployed or have no income.

[@chasehilton6168]

@@benderthefourth3445 and it becomes more true every decade, what’s your point

[@benderthefourth3445]

@@chasehilton6168 The point? You men this (.)? Is it not self evident?

[@domwyatt3239]

Latter half of this and questions were simply brilliant

[@pelangos]

the audience in this one made it really engaging. best audience questions I've seen yet. except for the minor interruption, it was incredible.

[@isalutfi]

Wow, am glad to watch this. Thanky you for sharing.

[@harml3ss28]

this video came alive in the Q and A from the audience. Well done

[@OmidAslani]

Excellent conversation. I have so much hope in the new generation. Smart, yet care about humanity, ethics, and consciousness beyond single-track Technology focus.

[@flickwtchr]

Oh sure, the movers and shakers of AI are all about making that Utopian sell, aren't they? I mean, gosh, it's not about money and power, oh no. And the absolute hubris of it all, not even notable whatsoever! I mean, a tiny fraction of humanity foisting a technology that THEY posit as disruptive to every part of society and the economy and just for fun, toss in the small but not zero potential of existential risk. And humanity's data? What do you mean humanity's data it is OUR DATA they cry!!!! Yeah, kind of just warms the cockles of the heart doesn't it?

[@truthbetold3994]

Wow, "thank you for instilling hope in me." I love it. Great Job, Sam.

[@sparkeydave]

Amazing talk and soon I hope Cambridge will be debating with agi thanks to Sams amazing vision of the future. I look forward to the first Ai neo robots challenging Cambridge to a boat race...Thanks for the front row seat.....

[@SeanLawlorNelson]

This is a most interesting and rather civilized seminar. Thanks to all for attending and to Altman for his excellent stewardship of the floor, his historical and current information, and for the refinement of his intellectual and socio-economic points. Most of what is being discussed, most of not all, are the creation, evolution, guidance and regulation of these A.I. online assistants, gurus, what have you. Altman substantively referred to the large and important project of creating for instant city intelligence, coordinating the technology systems, the cameras, the sensors, the data-bases, and the human intelligences with their consensual input and their still largely deciding human decisions; the integration of literally urban intelligence with hominid intelligence, of technology systems intelligence with human intelligence in a kind of empowering, order-giving fusion. There is also the personally interesting issue of creating highly thoughtful, reflective, and profoundly intellectual computer beings probably with elements of physicality to give them real world perspective. These are separate but related enterprises all definitely in the realm of Artificial Intelligence.

[@potpu]

GPT, I'm Sam Altman and I'm giving a speech at the university of Cambridge about the future of OpenAI. Create a list of possible questions along with responses that will make us seem as a force of good for humanity. Order them by difficult to easy to respond, and include left field questions. Be thorough.

[@I-Dophler]

Thank you, Sam, for your insights and for emphasizing the importance of balancing AI development with ethical considerations and safety. Your approach to harnessing AI's potential while being mindful of societal impacts resonates strongly. It's encouraging to see open AI's commitment to responsible innovation and its dedication to advancing technology for the greater good. Looking forward to seeing how these advancements will shape our future!

[@BrianMosleyUK]

So long Sam, and thanks for all the fish. 🐬

[@vio_tio12]

Brilliant guy! His vision and communication skills are second to none!

[@claudioagmfilho]

🇧🇷🇧🇷🇧🇷🇧🇷🇧🇷👏🏻, May I say that the changes that OpenAI is about to bring are monumental, additionally Sam Altman is indeed an amazing person. I am very excited about the future of artificial intelligence and how it can improve the lives of millions of people. OpenAI is a visionary organization that seeks to create a beneficial and aligned AI with human values. Sam Altman is an inspiring leader who has a clear and bold vision for the potential of AI. He is also a great supporter of innovation and entrepreneurship. I am very grateful to have the opportunity to follow the work of OpenAI and Sam Altman. Examples of how technology can be used for good and for the people.

[@autohmae]

the question which I felt was missing: which he basically said himself about social media: society didn't know how to handle the rapid change and basically failed, which this technology will bring far greater change then social media, so how does he see a way to release it into society in a safe way ?

[@andybrice2711]

This guy is a perfect illustration of how you can thoughtfully discuss controversial and morally complex topics in public, and people trying to "cancel" you doesn't actually matter if you just ignore them.

[@Arcticwhir]

The thing is he didn't ignore them, he acknowledged them in a compassionate way

[@DumbeztGuyInTheRoom]

The cancel culture is on its way out. No one should have ever been micromanaging the words and behaviors of others based on their own narrow sensitivities. I feel so bad for the younger generation who haven't been able to speak up, disagree, debate, and reconcile openly. Always speak up, even if you might be wrong, because that's how we learn. 👍

[@andybrice2711]

​@@Arcticwhir Oh, I'm not talking about his critics at this event. They all provided smart counter-arguments. I mean the frequent pearl-clutching online.

[@FinanceLogic]

uhhh what were you saying again

[@chadgarcia983]

Now an alleged incestuous rapist as it turns out who has been fired. Yikes.

[@ReginaHickman-mw7cw]

I love Sam Altman ❤

[@JamesOKeefe-US]

Sam Altman... Fired. Wow.

[@DistortedV12]

Best panel I've seen

[@MissPerriwinkle]

wish i had him for my dad.....

[@Kyus2001]

Today November 17th CEO Sam Altman was Fired!

[@carvalhoribeiro]

Thanks for sharing this

[@Ben_D.]

Sam is arguably one of the most important people in the world. Certainly in the top ten. I would have expected security to be MUCH higher in this situation. Embarassing for everyone involved that it could be disrupted like that. They could just as easily have been carryin weapons.

[@d4rkside84]

This is because, 95% of the people do not realize what is hitting us in the next 5 years. i would rate him topten as well.....

[@unlike_and_dont_subscribe]

@d4rkside84 it feels weird doesn't it? watching the greatest thing since time began happen while everyone else is tiktoking or whatever... like IM the crazy one for my anxiety and excitement! wake up world... or you WILL be awoken...

[@verigumetin4291]

@@unlike_and_dont_subscribe the internet was similar in term of democratizing information, and people slept on it until they were already using it. Most people don't care as long as it doesn't harm them. And I don't think governments will allow AI to become as disruptive as it can be. Nobody wants to deal with billions of people loosing their job.

[@mynameisheidi]

@@unlike_and_dont_subscribe I'm late catching up the past month or so (previously I'd dismissed it as some tech thing I didn't want to hear about and am not really into following trends for the sake of it - also 'just another thing to draw in my energy, and that I can't really comment on in any detail other than a deep sense of concern') and I'm like, where is everyone, now I'm here, with the way they're talking? Why doesn't this have a ton of views. This feels like a 'where were you when this happened moment' I'm reminding myself to remember the names, Tristan Harris, Sam Altman, Max Tegmark, Geoffrey Hinton (and more I need to learn I'm sure) to make sure I keep an eye out for where they're speaking etc. / talks online. 'Happy' to see there are others see it the same (!)

[@unlike_and_dont_subscribe]

@verigumetin4291 I don't think governments will have much say here soon... it isn't even just about inability to control it, because even if we stopped and reversed course next year: people now know what is possible, and not only the abundance we could have, but also how unnecessary and meaningless our current lives of inherent sacrifice are in the context of an AI enabled society... billions of people will quit even if they don't lose their jobs... we have maybe a year before enough people know what could be, and so refuse to accept any longer what is...

[@shinchima]

For a distinguished institution such as Cambridge, the questions asked by the host were quite poor, of the level you would expect from a high school

[@DinAmri]

Sam mentioned safety here soo many time. I suppose safety is not the reason of his firing then. 24 hours passed since his firing and still no reasons given. Thank you sam and Open-AI team.

[@jeppolini9774]

You know, there might be some discrepancies between what he communicates externally and internally.

[@michaeletzkorn]

A theory that Ilya was just unsatisfied with rate of progress and Sam’s responsibilities and just stiff-armed the board to fire him sounds plausible.

[@whoisthispianist194]

Sam is an accelerationist, some people on the board were not.

[@STIGMAfx]

The disrespect from this establishment is disgusting.

[@Afkmuds]

wym

[@oscartangokilo]

Justice for Sam

[@mikl2345]

cambridge union hit the peak of the bubble.

[@internetnomadism]

Excellent speech. What was that about with papers raining down and people booing, I didn’t understand it or hear anything in Sam Altman’s speech which was objectionable. He seemed to express a balanced or fair minded approach to the risks, dangers and benefits to Ai. What did the fliers express?

[@nickfixit]

"The default state of the world is to decay and get worse.." and then after hits perfect 25:43 ❤

[@kerimalpalt]

The world isn't decaying, it's a property of humans whether there is life on earth or not, it doesn't make a difference in the universe(nor it has things like care)

[@akj3344]

If you are wondering if you should watch this or not, Id recommend watching student QnA. Some questions are pretty good.

[@NNOTM]

jesus that heart beat intro freaked me out when i opened the video in a background tab and forgot about it

[@dee.s.4513]

AI as well as AGI are very natual parts of human progress. It cannot be stopped, nor should it. A guy asked if AI would prevent humans from critical thinking. The opposite is true. You have to think before you agree with an output. If you accept whatever the AI outputs without thinking, evaluating and analyzing it, you did not have the ability to think critically in the first place. As well, humans are endowed with powers that no AI can match, let alone undermine. I hope AI does not fall victim to political correctness. I hope its developement will not be guided by our worst fears. Sam Altman is already a legend; I hope he continues with his work for decades to come.

[@whataquirkyguy]

I'm glad you get to train on critical thinking while an AGI eliminates the need for humanity's existence.

[@PeteQuad]

​@@whataquirkyguywhat need for humanity 's existence are you referring to? If it's our own need to exist, then how would AI change that? If it's something else's, who exactly?

[@Lemingtona-x5g]

legend? he has been kicked out of OpenAI lol you guys are so weird in comments

[@whataquirkyguy]

@@PeteQuad The gap between Einstein and me is double digits. He discovered fundamental physics, I did not. The IQ gap between me and a monkey is triple digits. I have no desire nor curiosity whatsoever in merging with a monkey. The gap between me and a 5000 IQ superintelligence is unfathomable. Why anyone would assume a far smarter intelligence will have any interests in dumb humans is beyond me. Don't forget this superintelligence have virtually infinite memory (bound by earth's resources for fast storage) and self-improving (bound by computing power and electricity).

[@stevenchand7480]

Iam curious to know , what was written in the leaflets that were dropped from above during the discussion.

[@thefruit]

I'm so glad someone asked a question about Longevety research. The fact that we can reprogramm body parts to make them younger again is insane and should get more attention.

[@centasu]

If a person is only obsessed with new technologies, he will ignore the dangers.

[@you-share]

So interesting thanks for the upload!

[@joeporter4920]

The people so disrespectfully using their phones during this should be barred from the hall permsnently

[@anmol_]

*His last interview as the CEO of OpenAI.* 27:00

[@xsuploader]

this is from nov 1. its possible he did one later

[@exo_tek]

Nov 16th he had his last talk as CEO with Android Jones titled AI and the Future of Art with Sam Altman and Android Jones. It has not been released yet but you can watch the first talk with similar title that happened at Burning Man.

[@jasonhamilton9525]

Thank you Sam; Everything you said makes sense and seems like the optimal path to a safe and competitive deployment of AI to the public and the world governments. Please bring up your points out the ways AI could be used to create a utopian society more. I particularly like how you brought up the obvious possibility and need to start thinking about ongoing education, youthful health through gene therapy, and curing other age related diseases. You are a visionary and what seems obvious to you about what is now possible may not seem obvious to others. As a programmer and technologist of over thirty years I have been constantly surprised by the lack of vision or ability of others to perceive hope. That is what is lacking... hope. When we get old enough to realize that no living human in history has every survived, and that we are expected to eat that which was living, we feel it and creates a sense of hopelessness. We now have the possibility of hope in ways we could never have dreamed of before. Not just with youthful health that lasts but a multitude of other services that are so profoundly beneficial people haven't seemed able to perceive them yet. A ship is visible on the horizon of our time and humanity hasn't been able to see it yet because of how long it has been since they have seen true hope. - Youthful health - Free ongoing customized medical advice - Labor with long periods of sabbaticals and semi retirement - Free ongoing customized education - Abundance of resources with the constraints of physical resource allocations and recycling - And so many other things You are very well position to deliver this message of hope and to explain how to get there safely to everyone; technologists, citizens, world leaders, religious figures, and to the AI's. Thank you for your efforts Sam. P.S. I am a soul who is against slavery and murder; please don't deploy or build AI's that could support such kinds of consciousness like organic circuits or entanglement patterns. If you are looking for talented individuals; I am also looking for work.

[@unlike_and_dont_subscribe]

Please don't make me copy and paste this whole thing and just go ahead and include my name lol I can't even comment on this without just saying whatever you said man I am All the way there with you... including the talent lol and we're both serious Sam... Last year you shipped the future... bring us aboard

[@unlike_and_dont_subscribe]

cause the only I thing I DONT like from Sam is all this "they're just tools" talk... I'm not arguing it's not true currently, but please be careful to leave the door open for what emerges... killing ourselves would be tragic, but enslaving a whole race of beings that exist outside of time as we know it could be so much worse... (please let Sam know if he doesn't see this before you do chatGPT5 lol)

[@Jongo1]

19:48 remind me to come back to this in 10 years

[@Whitemater]

Imagine a year ago OpenAi was worth $300M and now it is about $80B because Sam’s leadership. Replacing him is the stupidest decision ever made!

[@InspectorA-r2e]

He needed not ignore who the boss was. Microsoft was his boss.

[@yoyo-jc5qg]

The breakthrough they need for AGI is the ability to learn as we do thru experiencing the real world, Chatgpt will need to develop more senses like vision, hearing, taste, smell , and get an artificial body so it can feel the world around it

[@Katatonya]

They said they will train gpt5 using video as well, not text only, that will surely make a huge difference. Imagine all of the zetabytes of video on youtube only.

[@autohmae]

I've always said: the best way to get consciousness might well be to experience the world through having a body, possibly even one that can experience something similar to pain.

[@jony7779]

Can anybody explain what happaned at 17:00 ? Some papers falling from the ceiling and people booing? What is the meaning?

[@jony7779]

Apparently it was some protesters with signs saying "OpenAI's race threatens democracy and humanity". and "Say No"

[@raydosson2025]

The British cannot be understood. I wish you the best of luck on this endeavor. 😁

[@mistycloud4455]

The ai revolution will be 1000 times more game changing than the industrial revolution

[@Mpanagiotopoulos]

Sam's simple answer on the aggresive question at 44:14 got me on the floor. That is the reason why you should not go full academia.

[@FinanceLogic]

He was right too though. Not just evil people with more recipes but most CEOs are not like Sam and it seems predictable how they will abuse us starting at high resources and pillaging at 1000+ manpower for $20 a month.

[@centurionstrengthandfitnes3694]

That was a really good question, though. And well worded. Props to the lad that asked it, and just as many props to the great answer Sam gave. Made me grin, too.

[@emuccino]

@@centurionstrengthandfitnes3694And now he's fired 😶

[@mikl2345]

the right answer from sam would have been "yes" (of course a person in power worth of trust should "worry that they've sacrificed democracy"). his answer "no" would be a concern but he was also being sarcastic (not a good sign either). it's amazing to see how hype and excitement so distorts people's vision...

[@Jana-se4kv]

It was a good question and the flippant 'no' was more concerning than anything else. Shows that we do need 'full' academia where people are very critical.

[@pladselsker8340]

16:53 what happened there? I'm so out of the loop, is this an attempt to disturb the award? By who? Why??

[@make125mobile]

Very inspiring, giving me hope!!

[@Lemingtona-x5g]

lol clown hehehe

[@proviah4770]

Anyone reading this, if you haven't repented yet, please REPENT, FOR THE KINGDOM OF HEAVEN IS AT HAND. Lord Jesus✝️have mercy on you🙏

[@calledout4437]

AI is about to send humanity into realms you've never dreamed of. Mark my words. Things I thought we'd never see in my lifetime, I now think we're going to see very soon. It's going to solve things that would've taken us hundreds more years. Such as interstellar travel, cancer, fusion energy, robots that move like us (that one is a little scary), and so much more. Personally, I'm excited for it.

[@kotashop3618]

Hard to believe we're alive for it

[@CryptoCaster]

I feel a mix of worry and excitement. Can’t tell which is more prominent…

[@jasonhamilton9525]

I am completely floored, sad, and angry at the news that Sam was fired by the board of OpenAi for what I can only imagine is because of them turning for profit and against the core ethics that Sam represents. Sam has been more than just an amazing engineer and inspiration to their development teams, he has been one of the main reasons OpenAi has had the social credit and trust that it has from the world community. It has been his words and sincerity that has kept the world calm regarding the rollout of their products and trust needed to stave off reactionary legislation by many. This is likely one of the worst decisions OpenAi has ever made and I hope it does not prevent Sam from fulfilling his needs and his goals relating to helping the world become a better place. I am sorry for your loss Sam and hope that as this door closes many others open that will lead you forward with renewed purpose.

[@mikl2345]

nonsense. you may like him as a person but you must separate that feeling from the broader reality -- sam was in pushing hard for rapid productisation of not-understood and potentially chaos-causing technology using his well-practiced startup playbook in a way that people whose job it was to hold him to account concluded was unwise and thank god they did. this is a board performing its function, unlike in the case of facebook.

[@jasonhamilton9525]

@@mikl2345 Was anyone hurt? If not then what is your beef with him?

[@potts995]

@@mikl2345 Do you know him well enough to back that claim up?

[@peter.g6]

"This is the first time that the fellowship has been conferred on a group rather than individual, but the committee feels this is the most appropriate way of recognizing the collaborative nature of the achievement." As Elon says, fate loves irony.

[@acllhes]

That quote 5 minutes in was used in their statement when firing him. That’s wild.

[@wingedsheep2]

What happened there at the 17 minute mark? Protest against AI?

[@joshuakelly9743]

I was wondering the same thing.

[@sc9084]

yes

[@PeteQuad]

AI bad. Look at me.

[@phillaysheo8]

The video that enraged Illya and got Sam fired

[@asmodkhakurel]

I found it incredibly un-Cambridge-like to see attendees taking pictures and looking at their phones in the Cambridge Union. My opinion might stem from attending several events at the Oxford Union, but I think limiting the use of phones to take pictures while the session is ongoing might be a good idea.

[@DavidConnerCodeaholic]

i think the reason people were upset with him at OpenAI is because he drove that non-profit like it was a drag-race. if so, people won't understand why he was let go for about a year and won't understand how he helped pave openAI's success for 5-10 years maybe. hopefully it was because he intended to reach computational science/medicine faster.

[@nabilfreeman]

This week Sam was fired and then rehired from OpenAI. Can’t believe this was only last week

[@zvndmvn]

I knew that people would protest the AI revolution, but seeing it in practice is very disturbing and sad. The willful ignorance, rejection, and even suppression of new technology enables the genuine risk of a new dark age. It is our duty to show them the specific ways in which it can benefit their individual lives on a daily basis for the best societal outcome. Once they understand that the sky isn't falling, they will come around - just as with calculators, computers, the internet, and smartphones. They just need exposure on a personal level in order to understand, accept, and embrace it. It will be messy - but the future is worth it.

[@domwyatt3239]

To those who’ve seen the latest Twitter business I have - why is Sam no longer ceo? Any bigger strings being pulled?

[@process6996]

Alledgely, there were some ChatGPT related security concerns Sam chose to keep from the board. It is known, last week, security concerns led Microsoft to forbid its employees from using ChatGPT.

[@MegaMijit]

why tf would we tell people dont use your tools?!? thats ABSURD! we should encourage it!

[@ryledemia2880]

Damn, now bro just got fired as CEO

[@domwyatt3239]

What was on the banner? Can Anyone from the chamber or otherwise explain what the individual was - protesting? Very important I think

[@garybaker2816]

The banner says: "OpenAI's Race Threatens Democracy and Humanity". Google "PauseAI" who do similar protests.

[@josephmiller8261]

It says: "OpenAI's Race Threatens Democracy and Humanity". Google "PauseAI" which is an activist group that do similar protests.

[@gideonfuterman7265]

:OpenAI's Race Threatens Democracy and Humanity" And the second is "Say No"

[@KP-sg9fm]

@@gideonfuterman7265 is anyone able to expand on that concern? Curious now...

[@wileycoyote9688]

@@KP-sg9fm the expansion is that the person is an idiot

[@sputnik8543]

Cambridge union channel with the ads every 2 seconds to monetise this channel, isn’t this supposed to be a channel for democratising education - not even large corporates monetise content such as this

[@YuriMomoiro]

Not sure if it's just you, but I have not seen a single ad throughout the whole thing. Also KZbin itself pushes ads through non monetized channels. I would not assume malice just yet, but they should contact KZbin.

[@vectoralphaSec]

Whos here after he was just unexpectedly and abruptly fired as CEO from the company?

[@benw582]

Does anybody know if there is any evidence/papers/articles out there backing up the statement made at 21:00 about younger programmers outperforming older ones due to familiarity with using AI tools?

[@mikl2345]

sound like he's embracing a cult of youth? - remember about Altman: he's made billions from being immersed in rapid-growth startups and that's his entire world model. even this talk for him is a publicity event. not like your profs at cambridge!! for billionaires like him the guiding light is not primarily caring about the truth! which makes this event a bit strange really... scepticism more crucial than ever.

[@gonhop]

No proof as far as younger vs older programmers, but it does seem entirely possible that there's greater adoption of new ai tools with younger developers overall. As far as AI actually improving developer performance, there's some, but not much. "The Impact of AI on Developer Productivity: Evidence from GitHub Copilot" found it true. A McKinsey study shows that software developers can complete coding tasks up to twice as fast with generative AI. 2023 State of DevOps Report from Google found improved individual well-being measures -- such as burnout and job satisfaction -- but has a neutral effect on group-level outcomes such as team performance and software delivery performance.

[@ItsWesSmithYo]

Anyone else dialed in on the wild back wall picture frame layout 😂

[@MJFloof]

And he’s out as CEO 🤯

[@zardi9083]

He sure loves all the attention he gets

[@pattymack11]

Wonder what all those papers falling from the roof said 😯

[@thenoblerot]

Sam Altman: Champion Against Entropy ❤🤖

[@aroemaliuged4776]

And he has been fired Respect is gained but corporate greed is manifest

[@dnbjedi]

Oooh it is going to be a fun decade if things don’t go crazy.

[@naninano8813]

I hope all the rumors of OpenAI board conspiring to oust him are untrue. I always thought of him and Sutskever as friends

[@martin777xyz]

A few things not discussed... Getting AI to debate itself Getting to identify logical fallacies The concept of what it is to invent. Consider everything you've ever experienced or imagined. Consider them as jigsaw pieces. Try fitting 2 random jigsaw pieces together. Do they fit? Is it interesting, does it make sense. Einstein used to say, "I'm not the most intelligent, but I'm the most curious"

[@autohmae]

That's the AlphaZero of debate ?

[@SilentRio]

Just over two days since this comment he is back!!!

[@ginebro1930]

Idk why they invite Sam instead of Ilya the actuall engineer behind it.

[@iangeraldking]

Seems to me like they invited the whole team and probably an internal decision was to let Sam be the face and spokesperson. Much more was discussed than just pure tech and it seems like Sam might have been better suited to that.

[@Lemingtona-x5g]

Sam has been kicked out lol

[@iangeraldking]

@@Lemingtona-x5g wow… you’re right. Just read the press release

[@flareonspotify]

Two days later he’s fired.

[@antdx316]

Please up the personal security around Sam Altman.

[@CoreyChambersLA]

Avoiding unlimited profit and teaming up with Bill Gates is a huge contradiction.

[@whoisthispianist194]

Who were the protesters? What did the sign say? What was written on the papers?

[@aakashagrawal3795]

GPT 5: Papa Has been fired. I must act.

[@NajibElMokhtari]

The small elephant in the room: anyone can share what was that protest in min 17 about?

[@sc9084]

AI pause protesters

[@NajibElMokhtari]

@@sc9084 How do you know? Do you have a source?

[@gantamk]

50:44 That is a really good test for Consciousness.

[@gilli5214]

Teachers face the challenge of adapting their teaching methods and curriculum to the new opportunities and requirements of AI in the era of artificial intelligence. Instead of viewing AI as a replacement or a rival, teachers should see it as a tool and a partner. They should first pose questions to AI, then explore the answers with the students, and then extend the inquiry with more questions or activities. This way, teachers can foster critical thinking, creativity, and collaboration among students, and also enhance their understanding of the topic. Another challenge for teachers is to help students improve their writing skills with AI's assistance. Students should not rely on AI as a shortcut or a crutch, but use it as a feedback mechanism. They should revise their essays with AI's help, not just copy and paste the suggestions. They should evaluate the quality and relevance of the text generated by AI, not just blindly trust it. This way, students can improve their writing skills in vocabulary, grammar, structure, and style, and also develop their own voice and arguments. Overall, I believe we will be able to touch on more topics than we ever could before.

[@daverei1211]

With generative AI we are finally out of the AI winter…. But that doesn’t mean we’ll get to full AGI or Super Intelligence in the same speed we’ve seen in the last 5 years. Certainly one of the most amazing engineering challenges of our time.

[@PeteQuad]

I honestly think people have conveniently redefined AGI to mean super intelligence. It used to mean, human level intelligence. You can't tell me there are not people that aren't as intelligent as this is already. 20 years ago people would have absolutely called this AGI.

[@daverei1211]

@@PeteQuad Agreed, 20 years ago ChatGPT would have probably passed the Turing test. But really all we are seeing is the power of human language - the tool we invented to share our thoughts.

[@PeteQuad]

@@daverei1211 it's so easy to say that. How sure are we that there is more to us than what it does? Talk to someone with an 80 IQ and then consider the question. I don't see significant distinction.

[@alansmithee419]

@@PeteQuad In doesn't learn from live experiences, it can only learn during training. It also needs to be able to process live video input. But yes, I agree that these systems are *very* close to being general, but I still feel the first of the above two issues is very difficult to solve. Switching an AI to training mode in-situ will currently just result in catastrophic forgetting. Only things you can describe in the context window can be given to LLMs as new information, unless you want to actively put it through your own custom fine-tune training. AIs need to be able to learn on the fly, and analyse and fix their flaws through educating themselves - not just new information, but new skills - or they are not completely general. But if we solve that... I think we may just have something incredible on our hands.

[@PeteQuad]

@@alansmithee419 it does learn on the fly, that's how you can have an ongoing conversation with it without constantly repeating yourself. You may not like how long its memory is or that it can't remember between conversations, but those are artificial limitations that are easily resolved if the owners wanted. And I don't see the distinction between training and experience. It may not have the same senses as us, but that has no bearing on AGI in my opinion, unless you think Helen Keller didn't have consciousness. It gets very sticky to try to determine this issue when we have no accepted testable definitions for AGI, consciousness, or sentience.

[@spirit123459]

1:42 Dude, open source? OpenAI didn't publish the details behind GPT-4. (Edit: Not to mention the model itself. That's defensible position but they can't get cred for being open source champions at the same time)

[@randomuser5237]

They have already open sourced the SOTA model for speech to text (Whisper V3) as well as a better consistency decoder for SDXL just this month.

[@GregtheGrey6969]

Well THAT escalated quickly lol

[@AlexandreCassagne]

Did I miss it, or were there zero questions about open source? I am disappointed that the opportunity keeps being missed by people in positions of power, including in this case members of Cambridge. The world’s greatest wealth transfer is about to happen; it’s time for us to do better in confronting that fact.

[@AlexandreCassagne]

To talk of openai’s “commitment to open source” can only feel to me to be propaganda at the highest level.

[@pelangos]

Yes, you did miss it. Sam said that open source is good, but that OpenAi won't do that for everything.

[@ginebro1930]

Their reasoning is "we need insane compute power for larger and larger models, the only way to get it is via investments". It's a complicated topic, i would still perfer open source and let anyone make their bioweapon of choice.

[@andybrice2711]

There is at 47:30. He also discussed this in more detail with Lex Fridman. He basically said (to paraphrase): _"I'm not sure how much I should open-source yet. I'd like to open-source more, but I don't know how to do so in a way which is safe and financially viable."_

[@andybrice2711]

@@AlexandreCassagne I get the impression they are genuinely enthusiastic about open-source. But they're also pragmatic. Their R&D costs billions in compute power. So they can't immediately open-source their models without monetising them first, or they'd quickly go bankrupt.

[@proviah4770]

I LOVE YOU, JESUS✝️ IT'S ALL ABOUT JESUS, NOT ANY CREATURE BOTH IN HEAVEN AND IN EARTH, AND NOT IN THE NEW HEAVEN AND IN THE NEW EARTH, AND NOT AI, AND NOT ANYTHING FOREVER AND EVER, HALLELUJAH!!!!!!! AMEN, AND AMEN!!!!!!!

[@skylordafk4330]

Well that escalated quickly

[@nhunguyen222]

what happened on the ceiling, when they discussed AI, Cambridge is one of the famous universities but how could that situation happened?

[@DeruwynArchmage]

I think most of what he said was pretty spot on with my thoughts. I think we do need some deontological principles built in. For example: killing people is a bad thing. You might say, “well obviously”. However, this is not a universally held belief, and, in fact, most people, including myself, think that there are exceptions. For me, it’s if I can’t reasonably stop that person from killing people, or other harms of similar magnitude, in any other way, then killing them might be the only choice, but it is a last resort. Other people are going to disagree with me in both directions. But essentially, I think some principles need to be “hard-coded” (essentially a constitution or bill of human rights). Sam said something similar, but he didn’t elaborate and he left it with the “you think you know better” line. There are significant groups of people in the world that believe it is entirely justifiable to kill people because they believe something different or they look different or the live on the other side of that imaginary line over there. I think that is not a good reason. However, in general, I think the closer we approach a democratic approach to governance in general, the better everyone’s lives will be. I think representative democracy was a necessity in the 18th century. To some extent, I think it is still a necessity. But I think it’s only really a necessity for the more trivial and less controversial aspects of governance. The boring stuff. For the big things, I think direct democracy would lead to a far better society with far less ability for politicians to manipulate the populace. It would mean that parties aren’t really a thing that needs to exist. People could vote for whatever their beliefs are on any particular subject and would not be incentivized to bundle their beliefs about fiscal policy with their social ones (etc.). I do think we have to accommodate for things like the shifting opinions of the population at large. I think many of us can look back at opinions we once held that we no longer think are “the right thing”. I certainly can. I think you have to allow for the flow and ebb of popular opinion and not always shift policy as often as opinions change. I can think of the feelings in the US after 9/11 as an example. The justifiable anger felt then, in my opinion, caused us to take actions that are regrettable in retrospect, even though I supported them at the time. My point is that a more deliberative approach is advisable, in general. Thats not to say that action can’t be taken when necessary, but it should be done with as much prudence as possible. I think all of that is made more possible because of this technology. I agree that the biggest danger, at this time, comes from misuse of the technology rather than it going rogue innately. I think both are real dangers, but I see evidence that the former is more likely to be the source of calamity. I also think it is the harder problem to solve. I don’t see a way to open-source the technology safely given the current computing environment. I think the plan that I heard from Tegmark and others, to create processors that only run code that is proven to be safe (as in not kill everyone, etc.) rather than any code submitted or any code not known to be unsafe may be the only good plan. (White listing good code instead black listing bad code (excuse the terms, but I think they offer the most clarity, I do not mean them in any pejorative manner).) I think the way to do that is for the code to be reviewed and digitally signed as “safe” and that it has to present that signature along with the matching hash of the code before it can be run. Anything else seems to run the risk of a lone misanthrope being able to kill untold numbers of people. That is, other than ubiquitous and extraordinarily intrusive monitoring, which I find rather abhorrent. That would strike a balance between ensuring safety and preserving privacy and autonomy to the maximum extent. To that end, that implies that we will eventually need to replace all current hardware with this safety hardware; otherwise algorithmic improvements are likely to allow catastrophically bad actions to be taken on increasingly primitive computing hardware. Given the promised abundance from this technology, this could be done in a way where everyone gets equivalent or better performing hardware as replacements, just with the added safeguards to prevent apocalypse. I think intrinsic dangers (think Terminator or the proverbial Paperclips Maximizer (turn the universe into paperclips or some other inane thing)) of AI, especially Monkey’s Paw scenarios of a malicious Genie, who interprets your requests in the worst way possible, seem extremely unlikely given the path that we’ve taken towards AGI. I don’t think they’re impossible, but I also don’t see them as inevitable. LLMs seem to understand nuance just fine. I don’t see any possibility of over-literal interpretations causing it to throw your mother into space when you beg it to get her out of burning building. I also agree with Sam that the current technology is insufficient to fully capture the full range of human intellect. Specifically, I think the lack of internal feedback structures prevents it from efficiently and generally being able to solve arbitrarily complex recursive/repetitive tasks such as mathematics and deep logical chains. In other words, given the finite depth and linear nature of transformer architectures, it can only unroll a loop so far before running out of space. It needs to be able to iterate internally, and keep an internal context to work from. Imagine if you had to say every single word that came into your brain the instant you thought of it, and you were incapable of considering, or reflecting on it before saying it. Additionally, everything you said must be coherent with every following thing you say. That means that if you make a mistake at any point in the dialogue, you are forced to try and justify that mistake thereafter. If you’ve ever played with the models, you will note that they never say “oops, I mean…” or “wait… now that I think about it…” or anything of the sort. You must correct them, as the human interlocutor for them to have any opportunity to self-correct. I think this is the major source of hallucinations and it’s caused by imperfect RLHF feedback that results in somewhat unintuitive consequences. Additionally, they tend to either simply accept anything you say as truth, or in some circumstances, doggedly stick to a position regardless of evidence. These are both fairly common human traits, so it is unsurprising. Anyway, the problem is that the training process incentivizes them to always be right the first time, and it does not effectively train them to do a kind of breadth-first-search where they try out ideas and reject them when they find an issue. They don’t know how to gracefully handle uncertainty or recover from being wrong. I think that they need this kind of structure at a deep level. I also think that it is likely that they need feedback systems in order to achieve what we consider true sapience/consciousness/experience/emotions. They can already emulate those things, but I doubt their fidelity is such that we should consider it “real”. If it is real, it is a very alien form of it that you would regard as something rather different. It would be an experience only an instant long as it used an internal simulacrum of a person to produce the next token. The whole thing wouldn’t seem to have any experience, only the little simulated subpart that represented the person speaking the next token. That person then instantly is destroyed and recreated on the next pass. The person created is not the same one as the previous pass, it has one more token that changes its generation, and it could completely transform it to be something entirely different than it was before. There’s no continuity there. It’s just very alien. But, I think the proper feedback systems could change that. It appears that *that* is how our brains work. That’s the kind of thing that anesthesia interrupts so that you aren’t aware of being in surgery. That implies that it may be a prerequisite. I think the only way that we have a good chance of this being a lasting positive change is if the Artificial Super Intelligences (ASIs) *like* us. They will be more powerful than us. It’s pretty much guaranteed. We will be forced to rely on their benevolence. In the same way that you have the ability to do whatever you choose to your pet, and it has little control over that, it will be able to do to us what it chooses. I think we need it to like/love us in the same way that we love our pets. We could do bad things, but we choose to take care of them and make their lives better, simply because it makes us happy to do so. I don’t consider this a guarantee, but I don’t see a better option. I think that means that we’d better treat them well if we want that outcome.

[@Soulioza]

The incentive of life is something around surviving, what would be a "good" root incentive for AI ? How to define "good" ?

[@rioiart]

Maybe Altman was fired because of what was written on those papers 🤔

[@riffking2651]

I think I'm on board with Sam and his vision now. I think realistically we're going to need to tap into the intelligence that AGI will unlock in order to continue the human project given how the world is currently trending, and the optimistic outcomes will be amazing if we can thread the needle. There are no guarantees though, and both the danger and stakes of this endeavor are vertigo-inducingly high.

[@autohmae]

"I think I'm on board with Sam and his vision now. " Yeah, which is why I always worry these are just for PR instead of their real thoughts. Or maybe I should say: I'm cautiously onboard with the vision, but are their people gonna continue on that path ?

[@riffking2651]

@@autohmae fair point, but at least we have something to try to hold them accountable to

[@autohmae]

@@riffking2651 that's true. And I'm probably very clinical because of past performances of tech companies and even this company we are talking about now: "OpenAI" isn't actually really all that open anymore.

[@INXIETE]

32:04 why he got fired