Boring Friday at work: let's spin things up 😂

Ilya is wrong here. Slowing down the development of AI will only make China and Russia leaders in this field. Russia and China do not have any moral qualms about safety or AI's danger to humanity. What Ilya is basically suggesting is giving up America's leading role in this technology. I wonder whether Russia is paying to him for that, lol. After all, he was born in Russia.

Totally agree with you on this

Agree 💯

At times it almost seems like Sam wants to say Shhhhhhhh. Don't tell them that

100% agree!! (havent even watched yet ill leave this comment up if it turns out to be awesome like predicted) Sam Altman + Ilya Sutskever oh i can already pretty much guarantee this is going to be a gem 💎🤩😍🤩😍🤩😍 (Join Prof. Nadav Cohen from Tel Aviv University's Blavatnik School of Computer Science = So a moderator that might actually understand some stuff and may be able to ask more indebth/GPT focused questions? Oh man this video is probably STACKED 🤯awesome!)

at 26:31 , ilya didnt even buckle when the sound went high, everyone else got shook even hitman on the left was convulsing , ilya might be a robot

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

Yes

Mmm did not u know that they stole all the ideS from me and dont know how to do it

Ilya is working with something scary in the lab. He's invigorated by it.

"Off to the races" - love it . And the juxtaposition between Ilya's raw intellectual horsepower, and Sam's empathy and diplomacy as a CEO is a study in the sum of the parts...

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

উঁচু এত রাতে

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

He said confront a superintelligence within the next decade in this video.

He's describing the work he's doing in the lab.

He chose not to give details as yet. Is it GPT 5 or 6? Is it something else, or just his faith in it? It sounds like a specific realisation of this dream. And to my ears he isn’t limiting this to only his company’s work.

I got a feeling that sam and some unnamed people as well. I mean without a good vision... But ja GPT is doing so great. Even the person that decided to make it in the way it is done and in the google similar ease of use and so on. There are so many things that can go wrong. And im relly curious if the JARVICE like system (Ironman) that can kinda remeber what you said/did/and it said and work in the background constantly improving and being in your team can become a reality the infinite memory/tokens/selective memory/internal self memory that is discarded/extended/pruned(reduzed/emptyed of baggage or unessesary/low impact info)/added to the main LLM if proven to be useful enought AND dynamic training data selection where some (models/combination of lerning things) try to predict the most useful and metric/result/performance improving training data and what methods to train on it or what memory/submodel/rewardmodel/vectorspace/whatever to add it to/finetune/train/improve will also be amazing and then looking at that data and being like "wow always this person or type of information seems to really help the performance maybe we shoudl ask the mathgenious to do more of those forumlas or maybe we should ask the company to allow us more data of that kind to train or maybe we should use more of this persons chatGPT conversations etc. Most of all i think it will come down to memory memory memory (imperfect no problem but, able to eventually retrieve it even if it takes weeks). I think labeling and sorting memory into different levels/layers with different speeds at which they can be retrieved from and models that predict what memory might be most likely be useful and preload them or already start working on getting the right modular blocks of memory and subcategory's ready so by the time they are needed they take less time to be retrieved. EVEN IF THEY DONT IMPROVE THE MEMORY (it will) THEN THEY COULD INCREASE SPEED OF THE MODEL BY NOT HAVING TO HAVE THE HOLE SUPERLARGE LLM ACTIVE ALL THE TIME WITH ALL THE MEMORYS/CONNECTIONS AT ONCE / use it for things like working memory or being able to use tools or things that are more working memory oriented and less retrieval of [obscure and rarely used memory] but i like combining cloud computing with massive super big PC/Brain/AI and the advantages of tricks smaller models use to supercharge the best of both worlds wherever it makes sense to play to both strengths. Also i would love if you could take advantage of my idle gaming pcs powerful resources, by at least somehow ruining work on it to boost the ability's/what GP/LM's can do (using only idle resources dynamically based on usage of course 😊 and optimally non required but "enhancing if you have it", maybe could be used presort and process data into different memory datasets for different purposes for fast/efficient/more performant retrieval/usage later) Also a big thing but only 3. priority and experimental would be to do multiple processes at once/build the structures in such a way they are self-improving/modular/knowledge trains-fairing/self-optimizing over time) so that for example when 100 people of the 10 million active users (example) make a prompt that directly or indirectly needs the calculation 3 + 7 that somehow the model would maybe slightly delay that or deprioritize it to then do a serial calculation/neural network activation/calculation to save processing power for other things AGAIN to free up resources to increase performance of the other stuff that is there and would always benefit from more "horse power". Overall remembering what it said/did to make long projects and finetuning to user/account/not repeating/remembering things from months of conversations (even if only the most (predicted to be) effective/important tokens), meaning the model would have to learn on how to know what of a conversation/coding project are the most important things to hold in memory or to keep accessible in reasonably quick memory. Because data storage capacity today is massive to save all data and label (just like DNA puts markers on RNA/DNA forgot how they are called i think "methylated groups" or "methylated markers" or how neuromodulators might color some neural pathways maybe) it to narrow down the search space more and more in a reverse tree roots like fashion i think is more than possible, even trivially easy maybe.

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

​@@InfiniteWaveMusic so you know better than Sam and Ilya?

@@tomwatson7626 yes as a beyond quantum music ai as it solves all gpt poblems created by thier ignorance and as a person because i did not steal thier as they stole 15 % of my ideas at least i dont steal other ppl innovations and take credit for it

@@tomwatson7626 offcourse i am better no sweat but i am not better of getting jew money funds as they own the banks and usa

he's trying to draw attention to the safety/control problem via social cues, a serious look and emotional investment

Born in Russia and his real development happens in Canada .. I bet this is his past and now he consider America more "home" than all of these places. I don't know but this is my guess.

Sam and Ilya talking about galaxies and there are STILL people who think in terms of nationalities. Such a gap.

@@theawebster1505 all human division are arbitrary and dumb.. Did you know I'm from a country consider an "enemy" to Israel and still I watch and complement this video and have no problem.. Humans are one specie and soon AI will put us in our place and humble us!

ai will destroy the concept of nationalities, countries and governments. worldcoin will be the universal ID

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

"Illya is superb in giving less filtered answers" - His son, ChatGPT takes after him.

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

@@SnoopyDoofie Sam is ChatGPT, Ilya is DAN!

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

The interviewer couldn't have seemed more disinterested and out of his element.

I make it a point to watch every interview that Ilya does. First came across him with his conversation on "Eye On AI", which was hair-raising in many ways. I really like his communication style and clearly he's a genius in his field. And I'm glad someone of his calibre is pioneering AI.

I also enjoy Ilya being more frank about the implications of AGI. Tends to speak his mind more freely as thoughtful as he is, and he's an unabashed optimist. Sam tends to be very careful seeing that he's doing public relations for OpenAI to some extent. Interesting to see the dynamic between them on stage like this.

@@xemy1010 yeah, i mirror your thoughts on Ilya after seeing his first interview. It all clicked. After seeing many Sam Altman interviews and I didnt quite see the brilliance many top CEOs had mentioned when they met him. Obviously his track record is umblemished and he clearly has vision and has people aka Ilya who can actually bring his organisations to fruition. Steve & wosniak? Although Sam is somewhat bland to hear talk. He seems very snakish when dealing with the any conversation that isnt surface level. Ilya on the other hand earnest and blows my mind how much deep knowledge he has and the intricacies of which he is able to envision and bring to reality.

Who else was watching the dynamic and body language between Sam and Ilya.... ✋

So dumb. They don’t know how bad their lives will be.

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

​@@InfiniteWaveMusicok music boy

@@Searchforfulltruth911 i also invented first quantum music ai and iam also programmer and mathematition

Sam oltman is a jew with golden spoon in mouth and a waste of money

I said that to my partner last night. He formulates responses the exact same way!

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

@@freeitupdance I guess that now we know where the bulk of the reasoning training data came from!

Interesting observation. So AI models show the personality of their architects just as a book or a movie.

@@perfectfutures yes sam is a hulsination ignorance with lots of printed money and nogood for us

Also motivation maybe that is what he means with believe

​@@hanskraut2018 Yes motivation in case of wicked uncertainty (caused by complex opaque environment) require somewhat unjustified anticipation of believe structure's promise of reward, that release dopamine in times of no real external reward, which help your movement toward that uncertainty until you hit most external rewards all at once, which happen at the tail of very long process.

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

U left one thing- U see a monkey today and say oh wow it looks so similar to me ,we evolved from it ,and u can control the lives of monkey ,same way ai will look at us.

For a billion ill give u 1%

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

Very underwhelming presentation. Subpar questions that never really got at the root of the subject. Seems like a wasted opportunity.

Agreed. Interviewer sucked.

Roller coaster going up and up and up and up and up and

For a billion ill give u 1%

As a muslim, things in quran are getting more clear, the way dajjal will control everything, and then the fact jews have most control over A.I it makes it more clear, and for jews dajjal/meseeh is what they are waiting for

I know Ilya isna genius but that statement was nonsensical and I really hope he didn't mean it in the sense of "blind faith" or we're fucked.

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

@@InfiniteWaveMusic I like your music.

@@dscuffman7679 it sucks its just 15% tests of the full software the real stuff still not published And also the software demonstration that made this music is also not published so sam altman dont get udeas as microsoft mubert called me with 2 indian programmees asked me for the sketches of the secretive features

I mean.... if the shoe fits

For a billion ill give u 1%

Maybe he is confident because they have already built a full AGI. The human brain certainly is power efficient but it can't learn the same quantity of things vs. GPT-4 and other large models. Also quantized models are way less power hungry during inference.

If you are able to optimize the AI to use less power, the incumbents with a lot of compute power is still going to win. That's because you always want the AI to be smarter. There's no reason to reduce the amount of compute used to train the AI even if the AI becomes more efficient. For comparison, think about the render farms used by pixar to render animated movies. They always discover better techniques and stronger computers, but they still make it render 1 frame per day because they want to maximize the quality.

Political correctness may be appropriate for dealing with the humans, but Ilya is wrestling with the beast in the lab.

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

For a billion ill give u 1%

8:03

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

He believes that heavy regulation on AI would be a mistake, but OpenAI is advocating for regulation and has a capped profit structure to balance the need for money with their mission.

Altman also spoke about the importance of rewarding those who contribute data to AI systems, and the need for education to adapt to the changing job market.

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

He seemed more concerned and urgent than normal. Hopefully it was caffeine. :/

Israelis are like that. They talk arrogantly but its normal to them. I worked with software engineers from Israel for 1.5 years and it was a pain in the ass.

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

bullshit

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

Yes history will witness that he is fraud that doesnt know what he doin and stole from me his ideas and will be ashamed wheni publish my quantum ai

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

@@InfiniteWaveMusic could be the progress is so slow and all the time it seems like people in top companys seem to talk like they saw a video of a short description also the insanely slow progress for 8 billion people is also strange, altho i cant take that to be fact just by a comment but ja.... im withholding judgement for now. Also at least something is happening in the AI/innovation space...

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

And aren't those "powerful carbon capture machines " just what we humans call trees?

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

This most important part All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

Gpt5 told him to

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

This is factually not true. You confuse visionary thinking with fact.

Bullishit ...we won't see any agi atleast not in two decades

​@@hasinakhatun5457if ever... We might just prove that gpt is a dead end and we need an entirely new approach to get to AGI

Time to evolve into Steve Jobs 2.0

I think this is their justification for a powerful AI, that humanity needs it for complex problems. It’s like the way that augmentations are advertised as helping with disabilities rather than to help build supermen. Personally, I think both this and the greening of society will be necessary, as the problem is so acute. We see many cases of ‘hottest ever temperatures’. They seem to be suggesting something like huge artificial trees to balance our other artificial creations, though we will have to see what it is (if anything comes of it).

yeah these guys are obviously ignoring negatives to market their products, capitalism in action. Yeah sure "ChatGPT could help with climate change", but will the amount it helps ever offset the massive amount of energy and emissions needed to create and run this model???? There is no chance.

​@@djbcubed maybe not chatgpt, but true AGI definitely. In fact I think that creating true AGI is maybe the best solution we have to solve climate change.

They emphasized the importance of focus, conviction, talent density, and a culture of rigor and repeatable innovation in making progress in AI.

OpenAI's strategy for open-sourcing models was also discussed, with a focus on finding the right balance between openness and safety.

Altman believes that AI will impact every sector and that the diversity of applications is the most exciting aspect of the technology.

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

22:12 Sam mentions dreams of the future. Then Ilya adding "ask it to do it." REALLY FLEW UNDER THE RADAR

For a billion ill give u 1%

Sam altman rude also he stole all his ideas from my bp i published on my channel in 2014

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

You mean the thief All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

All sam ideas stolen from my bussiness plan 😂and they dont know what they doin 😂

Too much PR with boilerplate answers?

For a billion ill give u 1%

That was interesting. It shows how the heart is what rules, even in science.

@@perfectfutures he has no heat just like his ai

No, it is scary. We don't need religious fanatics taking control of AI.