It’s the first interview where Emad is genuine. While Sam Altman is going around the world to lobby for GPT4, Emad can make models for big 😅companies and governments that don’t want to sign up giving away their secrets.
@@Eisenbison Right; go ahead and take credit of others' work (on the shoulder of giants), start companies, become a millionaire and appear on talk shows like Emad and we'll find out what you're made of.
"It's not about more data It's about better data" that is the key point in the convo for the future of AI. That is why LLM models are suddenly all open sourced, not because of altruism but because all the juicy data is behind private walls.
What kind of better data do you suggest? Are you saying that LLMs should be trained on a core corpus of carefully curated texts rather than just randomly scraping Reddit? How many of these tests would be a minimum and who would assess appropriacy? How of the data required do we already have? How much more will we need in the future?
@@christopherd.winnan8701 IMHO scraping the web and reddit can get you a decent broad knowledge LLM but if you want anything specific for PHD level domain knowledge you have to train smaller models on curated, organised high quality data, all of that data is behind private company walls or offline.
@@Zale370 Information is currently being hoarded under the guise of copyright and other restrictions. Copyright, for example, lasts for the life of the author plus an additional 70 years. This means that we will have to wait at least another couple of generations, before any of this material can be included in a truly open source LLM. Maybe it is time to draw a cut off for copyright at 2000 and everything else passes into the commons?
yeah, still life is not a point in time, the chart has multiple axis, time on X, profit on Y... is he gonna change like we all did? important things can't depend on one person, we need to act independently together to move the political agenda. Pharma controls puppet Drs. so... how is that going to be solved? how to get the consensus so the regular homer Simpson next door starts caring for himself and others?
Emad talks honestly and with an intention to teach viewers as many things as possible which is a breath of fresh air compared to other AI leaders(looking at you Sam Altman). More power to you Emad!
Unfortunately, he's about the most dishonest of them all. In investment pitches, he falsely claimed partnerships with the WHO and World Bank, made false statements about Stability AI's finances, failed to pay employees while his wife transferred large sums to her bank accounts, falsely claimed to have partly invented stable diffusion, and said he had a Master's from Oxford when he didn't. He's a former hedge fund manager now in Silicon Valley making grandiose predictions about "disrupting" industries he knows nothing about -- and he's about as unscrupulous as anyone should expect. It really sucks, but the people who are lauded as visionaries usually get that attention by self-promotion, mismanaging finances, and making things up.
@@douglasmyers-turnbull7136You copy-paste this stuff all around. Yes, Forbes wrote an extremely biased and dismissive piece on him. We know that. Now let's listen to what he actually has to say and then judge for ourselves, no? You don't think AI is going to disrupt economies across the board? Please, please back that up with something. Btw, funny you keep even trying to badmouth his wife... even the Forbes article acknowledged that the thousands (gasp!) of pounds they borrowed were all paid back in full.
@@douglasmyers-turnbull7136 Man. You're just regurgitating the Forbes hit piece with no criticical eye at all. Even the article, super biased against Mostaque as it was, acknowledged that the thousands (gasp!) he and his wife borrowed from the company were all paid back and accounted for. Please, there no need to spread disinformation. Listen to what the man is saying and then judge for yourself. Not based on some journalists personal vendetta
> Yes, Forbes wrote an extremely biased and dismissive piece on him. I guess Amazon and the UN were also out to get him. Unfortunately, I also read his response. He didn't dispute the claims. The only thing I'd side with him on is the Masters degree. He didn't technically have one (nor did he remotely earn one), but apparently Oxford just gives those out. > Now let's listen to what he actually has to say and then judge for ourselves, no? So I went back and watched the first part of the video. Right after the intro, we get this string of sentences around 7:15: "10% of people have a CYP450 mutation in their liver," ❌ The actual number is > 99.9% (counting CYPs that differ from a reference genome) "which means they metabolize drugs fast quicker" ❌ no particular CYP metabolizes all drugs "so if you metabolize codeine it turns into morphine" ❌It's literally supposed to. That's how codeine works. "or fentanyl kills you" ❌It's almost entirely metabolized to norfentanyl, which is inactive. "But that's a very basic genetic test that we give everyone 500 mg of the same thing" ✅/❌ I'll assume he meant we shouldn't give everyone the same dose. Pharmacogenetics can be useful, but dose escalations are a thing. "With my son, a microdose of 5 mg of clonazepam" ❌ That's an extremely high dose for a child, not a microdose. "which is used for anxiety disorder" ✅ Ok, sure. Sometimes. "allows him to sing" (I'll take his word for it) "But the standard dose is 1000 milligrams" WHAT???!! That's got to be a lethal dose. > You don't think AI is going to disrupt economies across the board? I never said that. I do think it will, but not all problems are currently amenable to AI, and many won't be for the foreseeable future. For example, his argument about pharmacogenetics is wrong. What's missing is data. Once we have that it's mostly a solved problem. The major metabolizing enzymes are known for most drugs, so we have causal links from allele -> rate. LLMs aren't much help without data. There lots of examples where LLMs can be used in medicine. It's almost like he picked a bad example because he never spoke with a pharmacologist, pharmacist, or AI expert. > Btw, funny you keep even trying to badmouth his wife No, I never did. I don't know anything about her. I think I joked she might not exist, which was a reference to Emad's truthfulness. > the thousands (gasp!) of pounds they borrowed were all paid back in full. Borrowing requires that you actually ask. His employees didn't agree to lend him money. The UK government considered intervening because of that. People treat him as a visionary. He's a former hedge fund manager that spews streams of obvious truths mixed with nonsense that sounds right. He's basically a stochastic parrot. I live in Silicon Valley, and I see a lot of these Elizabeth Holmes clones.
Emad is one of my favourite people in AI Love and learn endlessly from him , smart as Fxxx but his heart is in the right place , unlike many Great Interview
Not only has emad given me faith in the course of AI he's given me faith in humanity. What a legend 🙂 I too have utter faith in humanity. We are essentially good and it will shine through in the following years ☀️
Unfortunately he's an outright fraud. He's another George Santos / Elizabeth Holmes: Someone lying so continuously and with such extreme confidence that no one could think he's lying. But he never invented anything, isn't an AI expert, isn't partnering with WHO, didn't pay his employees on time, and even lied about receiving a Master's degree. He's a hedge fund manager who stole credit for others' work and lied to investors.
Emad really gets a lot of things right. National data sets make sense to me. I can't say that's the end-all, be-all, but at least as a minimum starting point, it could be extremely helpful for each sovereignty / nation / language center to have a "national data set." Then, each would need a blessed "national foundation model" (NFM) that could serve as a starting point for lots of spin-off models. It could also serve as a validation model. I also wonder if we should be working on fact-checking models. Then, each area / sovereignty could have a fact-checking model (or, if capable enough, the NFM can be that fact-checker). I'm wondering why building a foundational fact-checking model isn't already an active area of research. Maybe it is and I'm just unaware. Thanks for the interview.
This man is brilliant, I'm so happy to listen to him. I'm still trying to understand his take on OpenAI, he recognizes that they're goal is true general AI, but I think he's implying that this utopia that they seek, may become a nightmare, which Elon Musk also believes. The other great mind on this issue, LeCun (I may be misspelling his name), believes these large language models are inherently limited so there is nothing to worry about, despite the fear that their emergent behaviour elicits from Geoffrey Hinton, Sam Altman, Elon Musk, the theoretical physicist (I forget his name) among others. I agree these efforts should be open source because I fear that a few billionaires will have power over all of us long before a general intelligent agent becomes a reality.
Altman has previously talked about people receiving universal basic income and using their time to pursue more "meaningful activities like art and dance" where we presumably live in a sort of "leisure society". I'm sure even his collectivist utopian notion of re-defining the meaning and purpose of work would seem like a nightmare to many.
@@JumpDiffusion Well... he did just buy a $44Bn company and run it into the ground by with obviously bad choices, like indiscriminately firing thousands of employees, changing how users are verified, and scaring off investors. I mean... "great mind" is relative (and subjective).
Nope. He's literally just a fraud. He doesn't pay his employees, lied about his involvement and background, and told investors he was working with the UN and building a top-10 supercomputer. Elizabeth Holmes but somehow possibly worse.
The type of interview you need to listen more than once in case you missed something or to have a deeper understanding. Both individuals challenge your intelligence. Thanks You, from Ethiopia.
Emad, please go talk on the "All-in" podcast. They are very hungry for a guest like you, and you will have a very influential audience in terms of venture capital, where AI is going & where it needs to go. It has 351k subscribers, many of whom are in the tech field. The panel is mainly tech venture capital guys.
“If your are trying to align a more capable person, you are generally restricting its freedom! And it might not like that if it becomes aware!” - great talk
@@fantasticjames Sure. He claimed in investment pitches to have active partnerships with UNESCO, WHO, OECD, and the World Bank -- in fact, none of those organizations were working with him. He also called himself the "UN Covid AI lead" (while not working for/with the UN) and claimed he had a Master's degree from Oxford (he didn't). As for theft, the UK considered seizing his company's assets because he wasn't paying his employees. Basically, wage theft. Also, during that period, thousands of pounds went to his wife. I'd be surprised if this guy never ends up being charged with wire fraud at some point in his life. He's a former hedge fund manager who found himself at home among Silicon Valley snake oil VC. A bunch of "visionaries" who say they'll revolutionize healthcare, among other industries. About half of what he said about medicine here was wrong, (even about a drug he said his son takes). He says things with confidence that people assume it's true. He's another Elizabeth Holmes, Bankman-Fried, or Elon Musk.
@@douglasmyers-turnbull7136 Thank you for your reply. I respect your opinion & I agree with your point that he says things with complete confidence giving you the impression he's right. It is outstandingly convincing but when you consider his background in banking and finance we know that its been his job to be convincing for many years. Stebbings didn't challenge him too much & can be slightly sycophantic with his interviewees (which is not a criticism). His justifications were all monetary based, discussing $1bn payroll on deepmind and $150Bn of investment from Google to win this AI battle which is very much a hedge fund type rationale. All bachelor degrees from Oxford mature into Masters degrees so he could be telling the truth on this technicality. I was intrigued when he mentioned that journalists write badly about him but I haven't had the time to look into this but I figure its on the lines of the points you've made. I will look into it a little at some point.
He's onto the pulse of how this breaks. Training GPT's to do customer service jobs on the phone will be the first standardized breakthrough but it can be specialized and do a better job than insurance agents, financial advisors, etc. etc. Thats where its going to get hairy fast.
There is! The title is, "How to lie massively to everyone without even breaking eye contact, claim you invented things that you didn't, and be the absolute worst in human society."
In virtue's realm, let courtesy reside, A gentle grace that flows from heart's pure tide. Kind words and acts, like blossoms in the morn, Embrace the world, where love and peace are born. Integrity, a beacon shining bright, A steadfast guide through darkness and through light. With honesty, the soul stands tall and true, Unswayed by falsehoods, steadfast and imbued. In self-control, a sacred art is found, A mastery of self, where peace is crowned. Temper desires with wisdom's gentle hand, And find tranquility, like stillness of sand, Perseverance, a flame that burns within, Fanning the spirit, where dreams and hopes begin. Through trials and tribulations, ever strong, With steadfast courage, we shall journey along.
I loved this, and other podcast interviews I’ve been watching on the AI subject… I’d love to hear more about how this technology is going to affect my life… I’m on limited income and likely won’t be able to pay for a robot or even subscriptions to future apps that COULD help me… i live in a little, very economically poor mountain area… how will this technology be helping people in my region in the near future? There’s a lot of technical info i don’t understand… can someone help me understand better how this technology will be able to implement a utopian environment for me and people like me in this world…
Do you ever wonder why these CEO's of tech companies bother doing podcasts for channels with a few thousand subscribers? If you were the CEO of one of these super important new companies, wouldn't there be a million other things you should be doing instead, that would be higher priority?
If we're thinking purely from a selfish point of view then getting some fame might be the next step after wealth and power, and they might have to climb the ladder there too, e.g. start with small podcasts, then bigger ones, then a ted talk, etc. Of course it might just be a "sure why not" response to the request too.
It's about the quality of the subscribers and podcast. I'm not even familiar with this channel but your post echoes almost exactly what people said about high status people doing any podcasts 10 years ago, they thought podcasts were some fringe thing that would never really catch on. Doing bigger mega-popular podcasts would be great for reaching the public but If you want to attract investors, partnerships, top talent, and any early adopters there is tremendous value in reaching fringe audiences. I imagine that the listeners of small technology/VC/AI podcasts like this are heavily skewed towards exactly the kind of audience that AI startup execs want to reach.
The hundreds of thousands who are listening to him across all the podcasts he's done are his future customers, investors, employees and old & new (like myself) fans.
Real time Natural occurring language is the best dataset for training a language model. You need to be able to ask questions when you learn and with random internet data the user is not there to answer your questions
The 'democratization' of anything always ends up making a small number of people disproportionately rich, and a large number of people poorer, increasing wealth inequality. Not everyone wants or needs MORE MORE MORE, nor should they be forced to keep up.
The forthcoming investment into different AI projects represents different ideas as to what may, or may not, be valuable. No one possesses a monopoly on the answer. Waste? Nay, necessary funding of experimentation.
I agree with him, we need national data bases so that we keep control of OUR data, and prevent these VC leeches and others from getting their filthy hands on it and syphoning off any gains to themselves.
National Datasets should not fall into the same trap as current myopic Localisation by State/Country of origin. It should be that each individual is a mobile representative of their own origin(s). Identity can be associated with the personal AI (Avatar) we will get, which in turn on the network-facing side, collaborates with its peers, not just Inter- speciel but Intra-speciel so the 'biosphere' unites (some time later) The bee flies through the window and you can say "Want some Honey? :) )
While national datasets will provide better results because of local context - doesn't it have an issue of creating knowledge echo chambers or especially catastrophic in countries where governments try to control mass communications?
That's one hypothetical example of a massive amount of issues coming our way... YES there will be issues. K I bet knowledge echo chambers will be everywhere, national models or not. BUT there is no stopping the train, so better try to come up with scenarios to benefit humanity
57:18 unfortunately this all means people will generally become illiterate if you can not write on your own, just like how most people start losing math skills once they depend on a calculator
Interesting... Mostaque has said before he is super Bullish on Apple and that it is the biggest AI company in the world no one is talking about because of the unused neural engine in the chipset.
It is not coming like a train…it is a hyper sonic cruise missile…we are about 2-3 years away from a global catastrophe which will take us back into the stone age
Quite a few nuggets of what sounds like "crazy" predictions but they were my position before even watching this: * Coding will not be a thing in just a few short years * Money might not be a thing anymore * You could make your own armies of human-level AI soon * OpenAI only cares about bringing AGI and utopia * Many jobs are going to vanish soon and in a short duration, to recover you must adapt quickly
you missed. it litrally says "AGI and utopia" and you only read "utopia". Utopia is good. But AGI "and" utopia sounds really, really.. i cant even @@alpha0xide9
56:50 I've got a really solid plan for AGI Alignment so I'll be looking to talk to #EmadMostaque very soon to get that Billion in funding. It's like nothing anyone else is considering, as far as I can tell, and it will be a whole lot of fun that brings the world together on an internal Alignment journey for humanity itself.
I doubt it. Usually people commenting in RUclips claiming they have a solution also have a case of Dunning Krueger. The real talents are rarely commenting on RUclips videos and claiming they have a solution to strangers. Instead, they are working in teams of other exceptionally gifted and talented people or also researching or in academia. If you’ve thought about a solution, they probably have already thought of it and discarded it. With the amount of actually intelligent people working on AI right now, there are probably dozens of ideas that are better than yours.
@@douglasmyers-turnbull7136 is it just a conspiracy theory "out there" or is it something that is not known to the general public who really think Emad is a clever and creative guy???
CEO stability AI Emad Mostaque is for humanity, transparently and open. CEO OpenAI Altman is for-profit closed Source. I think the world need a national data source for the benefit that the people us what he al mention. His quote what I saw in a other interview: "AI by the people, for the people".
JESUS✝️REIGNS FOREVER MORE, HALLELUJAH!!!!!!! CELEBRATE AND GLORIFY JESUS EVERY DAY AND FOREVER MORE, HALLELUJAH!!!!!!! AMEN. JESUS✝️IS THE ONLY HOPE FOR HUMANITY!!!!!!! I LOVE YOU JESUS✝️ REPENT, FOR THE KINGDOM OF HEAVEN IS AT HAND. Lord Jesus✝️have mercy on you🙏
No, it's really, really not. He's been exposed as a fraud with no apparent moral compass. He lies to investors, reporters, and the public about pretty much everything, including that he's partnering with the WHO and even that he has a Master's degree (he doesn't). Elizabeth Holmes is probably more trustworthy, and her lies could have caused people to die.
„The investor in Stable Diffusion“ would probably the more appropriate monicker - not „the creator“. And that shouldn’t discount him - you can’t create without investment
There’s a lot of coherence in what Emad says, but at the same time it seems a bit arrogant to imagine that the next years are so predictable. At this point in time, with the exponential growth of science and knowledge, it’s impossible to predict what’s going to happen even in the next month…
43:25 ….Not many people were writing assembly 20 years ago. There were no libraries? I call BS ;) I was writing Haskell and Python 20 years ago. C++ had boost over 20 years ago. Emad is a fast talker and he talks BS sometimes ;)
Good God no. He can't even pay his employees on time. He's a massive fraud and habitual liar who's going to crash to the floor burning like Elizabeth Holmes did.
Why would anyone build an AI publishing business when your own personal AI on your device does all your personal news for you? All we need is AI reporters to feed JSON to the hive to propagate. Imagine, unbiased embodied AI journalists holding a flesh sack human politician to account. What a weird world.
![Felix "Unfair" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/Oswujxm2Ag0/mqdefault.jpg)
This is right now the Internet’s MOST UNDERRATED interview. Brilliant, mind-bending, illuminating!!!
Can’t agree more!
Agree
It’s the first interview where Emad is genuine. While Sam Altman is going around the world to lobby for GPT4, Emad can make models for big 😅companies and governments that don’t want to sign up giving away their secrets.
Sounds about right
From the sound of things, he's not nearly as smart as he makes himself out to be and has a history of taking credit for other people's work.
Emad is simply brilliant. Great interview.
Maybe there’s hope for humanity ‘cos there’s no way a computer will ever be as smart as this dude 🤣
Ethics rather than intelligence I think
He's actually not all that smart, and has a history of taking credit for other people's work and exaggerating his own achievements.
Intelligence does not automatically/necessarily equal desire to dominate/conquer.
@@Eisenbison Right; go ahead and take credit of others' work (on the shoulder of giants), start companies, become a millionaire and appear on talk shows like Emad and we'll find out what you're made of.
You are certainly wrong start to stop
"It's not about more data It's about better data" that is the key point in the convo for the future of AI. That is why LLM models are suddenly all open sourced, not because of altruism but because all the juicy data is behind private walls.
Make sense but lets see
That’s an interesting point of view 🤔
What kind of better data do you suggest?
Are you saying that LLMs should be trained on a core corpus of carefully curated texts rather than just randomly scraping Reddit? How many of these tests would be a minimum and who would assess appropriacy?
How of the data required do we already have?
How much more will we need in the future?
@@christopherd.winnan8701 IMHO scraping the web and reddit can get you a decent broad knowledge LLM but if you want anything specific for PHD level domain knowledge you have to train smaller models on curated, organised high quality data, all of that data is behind private company walls or offline.
@@Zale370 Information is currently being hoarded under the guise of copyright and other restrictions. Copyright, for example, lasts for the life of the author plus an additional 70 years. This means that we will have to wait at least another couple of generations, before any of this material can be included in a truly open source LLM.
Maybe it is time to draw a cut off for copyright at 2000 and everything else passes into the commons?
Among the AI leaders, Emad seems the most 'human' and relatable.
yeah, still life is not a point in time, the chart has multiple axis, time on X, profit on Y... is he gonna change like we all did? important things can't depend on one person, we need to act independently together to move the political agenda. Pharma controls puppet Drs. so... how is that going to be solved? how to get the consensus so the regular homer Simpson next door starts caring for himself and others?
Agreed. The more I listen to him, the more I respect where he is coming from
Every time I heard to Mostaque's perspectives I come with new (progressively) more powerful ideas.
Emad talks honestly and with an intention to teach viewers as many things as possible which is a breath of fresh air compared to other AI leaders(looking at you Sam Altman). More power to you Emad!
Unfortunately, he's about the most dishonest of them all. In investment pitches, he falsely claimed partnerships with the WHO and World Bank, made false statements about Stability AI's finances, failed to pay employees while his wife transferred large sums to her bank accounts, falsely claimed to have partly invented stable diffusion, and said he had a Master's from Oxford when he didn't.
He's a former hedge fund manager now in Silicon Valley making grandiose predictions about "disrupting" industries he knows nothing about -- and he's about as unscrupulous as anyone should expect. It really sucks, but the people who are lauded as visionaries usually get that attention by self-promotion, mismanaging finances, and making things up.
@@douglasmyers-turnbull7136You copy-paste this stuff all around. Yes, Forbes wrote an extremely biased and dismissive piece on him. We know that. Now let's listen to what he actually has to say and then judge for ourselves, no?
You don't think AI is going to disrupt economies across the board? Please, please back that up with something.
Btw, funny you keep even trying to badmouth his wife... even the Forbes article acknowledged that the thousands (gasp!) of pounds they borrowed were all paid back in full.
@@douglasmyers-turnbull7136 Man. You're just regurgitating the Forbes hit piece with no criticical eye at all.
Even the article, super biased against Mostaque as it was, acknowledged that the thousands (gasp!) he and his wife borrowed from the company were all paid back and accounted for.
Please, there no need to spread disinformation. Listen to what the man is saying and then judge for yourself. Not based on some journalists personal vendetta
> Yes, Forbes wrote an extremely biased and dismissive piece on him.
I guess Amazon and the UN were also out to get him.
Unfortunately, I also read his response. He didn't dispute the claims. The only thing I'd side with him on is the Masters degree. He didn't technically have one (nor did he remotely earn one), but apparently Oxford just gives those out.
> Now let's listen to what he actually has to say and then judge for ourselves, no?
So I went back and watched the first part of the video. Right after the intro, we get this string of sentences around 7:15:
"10% of people have a CYP450 mutation in their liver," ❌ The actual number is > 99.9% (counting CYPs that differ from a reference genome)
"which means they metabolize drugs fast quicker" ❌ no particular CYP metabolizes all drugs
"so if you metabolize codeine it turns into morphine" ❌It's literally supposed to. That's how codeine works.
"or fentanyl kills you" ❌It's almost entirely metabolized to norfentanyl, which is inactive.
"But that's a very basic genetic test that we give everyone 500 mg of the same thing" ✅/❌ I'll assume he meant we shouldn't give everyone the same dose. Pharmacogenetics can be useful, but dose escalations are a thing.
"With my son, a microdose of 5 mg of clonazepam" ❌ That's an extremely high dose for a child, not a microdose.
"which is used for anxiety disorder" ✅ Ok, sure. Sometimes.
"allows him to sing" (I'll take his word for it)
"But the standard dose is 1000 milligrams" WHAT???!! That's got to be a lethal dose.
> You don't think AI is going to disrupt economies across the board?
I never said that. I do think it will, but not all problems are currently amenable to AI, and many won't be for the foreseeable future.
For example, his argument about pharmacogenetics is wrong. What's missing is data. Once we have that it's mostly a solved problem. The major metabolizing enzymes are known for most drugs, so we have causal links from allele -> rate. LLMs aren't much help without data.
There lots of examples where LLMs can be used in medicine. It's almost like he picked a bad example because he never spoke with a pharmacologist, pharmacist, or AI expert.
> Btw, funny you keep even trying to badmouth his wife
No, I never did. I don't know anything about her. I think I joked she might not exist, which was a reference to Emad's truthfulness.
> the thousands (gasp!) of pounds they borrowed were all paid back in full.
Borrowing requires that you actually ask. His employees didn't agree to lend him money. The UK government considered intervening because of that.
People treat him as a visionary. He's a former hedge fund manager that spews streams of obvious truths mixed with nonsense that sounds right. He's basically a stochastic parrot.
I live in Silicon Valley, and I see a lot of these Elizabeth Holmes clones.
That guy is awesome
Emad Mostaque is an international treasure
Emad is one of my favourite people in AI Love and learn endlessly from him , smart as Fxxx but his heart is in the right place , unlike many Great Interview
Wow, amazing interview, amazing guy.. so much insight and intelligence there...
What a informative, relevant and forward thinking interview, great work to both!
LOVED this. One of Emad's best chats. Id love to work at Stability
Don't. The founder is a fraud who doesn't pay his employees on time and steals things to slap his name on them.
Not only has emad given me faith in the course of AI he's given me faith in humanity. What a legend 🙂 I too have utter faith in humanity. We are essentially good and it will shine through in the following years ☀️
Unfortunately he's an outright fraud. He's another George Santos / Elizabeth Holmes: Someone lying so continuously and with such extreme confidence that no one could think he's lying. But he never invented anything, isn't an AI expert, isn't partnering with WHO, didn't pay his employees on time, and even lied about receiving a Master's degree. He's a hedge fund manager who stole credit for others' work and lied to investors.
Emad really gets a lot of things right. National data sets make sense to me. I can't say that's the end-all, be-all, but at least as a minimum starting point, it could be extremely helpful for each sovereignty / nation / language center to have a "national data set." Then, each would need a blessed "national foundation model" (NFM) that could serve as a starting point for lots of spin-off models. It could also serve as a validation model. I also wonder if we should be working on fact-checking models. Then, each area / sovereignty could have a fact-checking model (or, if capable enough, the NFM can be that fact-checker). I'm wondering why building a foundational fact-checking model isn't already an active area of research. Maybe it is and I'm just unaware. Thanks for the interview.
Spoken like a true Marxist
@@pampypants871 huh? I sense a non-sequitur.
@@pampypants871You have no clue what he is talking about 😂
great podcast.. for the first time I had to slowdown the speed increase of increase it.. 😊
Best conversation on open AI on RUclips
This man is brilliant, I'm so happy to listen to him. I'm still trying to understand his take on OpenAI, he recognizes that they're goal is true general AI, but I think he's implying that this utopia that they seek, may become a nightmare, which Elon Musk also believes. The other great mind on this issue, LeCun (I may be misspelling his name), believes these large language models are inherently limited so there is nothing to worry about, despite the fear that their emergent behaviour elicits from Geoffrey Hinton, Sam Altman, Elon Musk, the theoretical physicist (I forget his name) among others. I agree these efforts should be open source because I fear that a few billionaires will have power over all of us long before a general intelligent agent becomes a reality.
Imagine thinking Elon Musk a great mind. Lol.
LLM’s are already commoditized, thanks to the Facebook leak.
Altman has previously talked about people receiving universal basic income and using their time to pursue more "meaningful activities like art and dance" where we presumably live in a sort of "leisure society". I'm sure even his collectivist utopian notion of re-defining the meaning and purpose of work would seem like a nightmare to many.
@@cholst1 Elon Musk is not a great mind? Ok, son....
@@JumpDiffusion Well... he did just buy a $44Bn company and run it into the ground by with obviously bad choices, like indiscriminately firing thousands of employees, changing how users are verified, and scaring off investors. I mean... "great mind" is relative (and subjective).
What are the 5 companies?
Great episode. Appreciate your work to put this together
Absolutely love the podcast, but a bit of due diligence on shadiness of guest could be cool. Again Best podcast out there
excellent interview
Amazing intelligent interview and I love it & understood every word.
this guy is a breath of fresh air
Nope. He's literally just a fraud. He doesn't pay his employees, lied about his involvement and background, and told investors he was working with the UN and building a top-10 supercomputer. Elizabeth Holmes but somehow possibly worse.
Fantastic episode
The type of interview you need to listen more than once in case you missed something or to have a deeper understanding. Both individuals challenge your intelligence. Thanks You, from Ethiopia.
Emad, please go talk on the "All-in" podcast. They are very hungry for a guest like you, and you will have a very influential audience in terms of venture capital, where AI is going & where it needs to go. It has 351k subscribers, many of whom are in the tech field. The panel is mainly tech venture capital guys.
"It should be open by default. The value is not in any propriety models or data." Amen to that!
“If your are trying to align a more capable person, you are generally restricting its freedom! And it might not like that if it becomes aware!” - great talk
openness is essential to intelligence....without openness, learning & teaching stops
I love this guy
That's unfortunate because he's a fraud.
insightful interview !
Incredible interview. Thank you so much for this. Are you open for suggestions on new guests?
this was great. i feel smarter just watching this. didnt think i could get more into ai but i feel like running thru a wall rn
Brilliant Episode! Thanks for sharing!
This is a very intelligent guy.
Well, he's good at stealing things and lying point-blank without breaking a sweat. I guess that's a kind of intelligence.
@@douglasmyers-turnbull7136 Please can you provide examples of stealing and lying.
@@fantasticjames Sure. He claimed in investment pitches to have active partnerships with UNESCO, WHO, OECD, and the World Bank -- in fact, none of those organizations were working with him. He also called himself the "UN Covid AI lead" (while not working for/with the UN) and claimed he had a Master's degree from Oxford (he didn't).
As for theft, the UK considered seizing his company's assets because he wasn't paying his employees. Basically, wage theft. Also, during that period, thousands of pounds went to his wife. I'd be surprised if this guy never ends up being charged with wire fraud at some point in his life.
He's a former hedge fund manager who found himself at home among Silicon Valley snake oil VC. A bunch of "visionaries" who say they'll revolutionize healthcare, among other industries. About half of what he said about medicine here was wrong, (even about a drug he said his son takes). He says things with confidence that people assume it's true. He's another Elizabeth Holmes, Bankman-Fried, or Elon Musk.
@@douglasmyers-turnbull7136 Thank you for your reply. I respect your opinion & I agree with your point that he says things with complete confidence giving you the impression he's right. It is outstandingly convincing but when you consider his background in banking and finance we know that its been his job to be convincing for many years. Stebbings didn't challenge him too much & can be slightly sycophantic with his interviewees (which is not a criticism). His justifications were all monetary based, discussing $1bn payroll on deepmind and $150Bn of investment from Google to win this AI battle which is very much a hedge fund type rationale. All bachelor degrees from Oxford mature into Masters degrees so he could be telling the truth on this technicality. I was intrigued when he mentioned that journalists write badly about him but I haven't had the time to look into this but I figure its on the lines of the points you've made. I will look into it a little at some point.
@@douglasmyers-turnbull7136You also read the Forbes hit piece? Great.
One of the few times when it pays to go 0.75x 😅😊 what a man, what a mind!
Great interview!
love from 🇧🇩
Nice channel dude. 20VC
He's onto the pulse of how this breaks. Training GPT's to do customer service jobs on the phone will be the first standardized breakthrough but it can be specialized and do a better job than insurance agents, financial advisors, etc. etc. Thats where its going to get hairy fast.
Dude, what will the 10 thousands of Philipino call center/customer service workers do then? Alot of livelihoods will be affected by AI.
I feel like this guy actually cares about humans
every medias should relay this one!
tl;dr; what are five companies and why?
Is there a course by Emad Mostaque?
There is! The title is, "How to lie massively to everyone without even breaking eye contact, claim you invented things that you didn't, and be the absolute worst in human society."
Is a personal AI now available to the public?
Depends on what you mean by it
@@alpha0xide9 I mean one like Chat GPT which I can have to read books by the authors I chose and biases I like etc
In virtue's realm, let courtesy reside,
A gentle grace that flows from heart's pure tide.
Kind words and acts, like blossoms in the morn,
Embrace the world, where love and peace are born.
Integrity, a beacon shining bright,
A steadfast guide through darkness and through light.
With honesty, the soul stands tall and true,
Unswayed by falsehoods, steadfast and imbued.
In self-control, a sacred art is found,
A mastery of self, where peace is crowned.
Temper desires with wisdom's gentle hand,
And find tranquility, like stillness of sand,
Perseverance, a flame that burns within,
Fanning the spirit, where dreams and hopes begin.
Through trials and tribulations, ever strong,
With steadfast courage, we shall journey along.
Pray tell me whence
I loved this, and other podcast interviews I’ve been watching on the AI subject… I’d love to hear more about how this technology is going to affect my life… I’m on limited income and likely won’t be able to pay for a robot or even subscriptions to future apps that COULD help me… i live in a little, very economically poor mountain area… how will this technology be helping people in my region in the near future? There’s a lot of technical info i don’t understand… can someone help me understand better how this technology will be able to implement a utopian environment for me and people like me in this world…
He's very intelligent. And sharp. 54:00 he just confirmed what George Hotz said on Lex, GPT-4 has 220billion parameters. Amazing.
Dude why do u only have 20k subs? You should have more.
❤$RNDR going to pop @ Jun 5 Apple metaverse conference 😊
i agree!
Do you ever wonder why these CEO's of tech companies bother doing podcasts for channels with a few thousand subscribers? If you were the CEO of one of these super important new companies, wouldn't there be a million other things you should be doing instead, that would be higher priority?
If we're thinking purely from a selfish point of view then getting some fame might be the next step after wealth and power, and they might have to climb the ladder there too, e.g. start with small podcasts, then bigger ones, then a ted talk, etc. Of course it might just be a "sure why not" response to the request too.
Both very good points
It's about the quality of the subscribers and podcast. I'm not even familiar with this channel but your post echoes almost exactly what people said about high status people doing any podcasts 10 years ago, they thought podcasts were some fringe thing that would never really catch on. Doing bigger mega-popular podcasts would be great for reaching the public but If you want to attract investors, partnerships, top talent, and any early adopters there is tremendous value in reaching fringe audiences. I imagine that the listeners of small technology/VC/AI podcasts like this are heavily skewed towards exactly the kind of audience that AI startup execs want to reach.
The hundreds of thousands who are listening to him across all the podcasts he's done are his future customers, investors, employees and old & new (like myself) fans.
It's still a niche thing
Is there a course to upgrade a programmer to an ai engineer?
Interesting dude
ufff this was great
1:09:04 Humans might trust Google Maps. But they don't trust Google.
Real time Natural occurring language is the best dataset for training a language model. You need to be able to ask questions when you learn and with random internet data the user is not there to answer your questions
The 'democratization' of anything always ends up making a small number of people disproportionately rich, and a large number of people poorer, increasing wealth inequality.
Not everyone wants or needs MORE MORE MORE, nor should they be forced to keep up.
and there's yann LeCun's new design that really needs to be paid attention to
The forthcoming investment into different AI projects represents different ideas as to what may, or may not, be valuable. No one possesses a monopoly on the answer. Waste? Nay, necessary funding of experimentation.
I agree with him, we need national data bases so that we keep control of OUR data, and prevent these VC leeches and others from getting their filthy hands on it and syphoning off any gains to themselves.
We need mass treatment in the veterinary world for animals
National Datasets should not fall into the same trap as current myopic Localisation by State/Country of origin. It should be that each individual is a mobile representative of their own origin(s). Identity can be associated with the personal AI (Avatar) we will get, which in turn on the network-facing side, collaborates with its peers, not just Inter- speciel but Intra-speciel so the 'biosphere' unites (some time later) The bee flies through the window and you can say "Want some Honey? :) )
While national datasets will provide better results because of local context - doesn't it have an issue of creating knowledge echo chambers or especially catastrophic in countries where governments try to control mass communications?
That's one hypothetical example of a massive amount of issues coming our way... YES there will be issues. K
I bet knowledge echo chambers will be everywhere, national models or not. BUT there is no stopping the train, so better try to come up with scenarios to benefit humanity
57:18 unfortunately this all means people will generally become illiterate if you can not write on your own, just like how most people start losing math skills once they depend on a calculator
And the bee replies, " eh, no I make it. Flowers?". "Balcony that way..."
Any reason why your trying to burn 20VC onto my telly.?
How does this have so few views?
Fab Videos
You know when you think fucking hedge fund. This is not that guy this is my guy peace and love
Interesting... Mostaque has said before he is super Bullish on Apple and that it is the biggest AI company in the world no one is talking about because of the unused neural engine in the chipset.
Whatever you do with ai pls dont take away my pornhub!
Kvetching is valuable criticism in aggregate. So people complain matter.
Articulate …. Nice!
It is not coming like a train…it is a hyper sonic cruise missile…we are about 2-3 years away from a global catastrophe which will take us back into the stone age
Quite a few nuggets of what sounds like "crazy" predictions but they were my position before even watching this:
* Coding will not be a thing in just a few short years
* Money might not be a thing anymore
* You could make your own armies of human-level AI soon
* OpenAI only cares about bringing AGI and utopia
* Many jobs are going to vanish soon and in a short duration, to recover you must adapt quickly
money, including all forms of it, will be irrelevant.
@@clusterstage all about elements and energy
"Only cares about bringing utopia" is a pretty strange statement
you missed. it litrally says "AGI and utopia" and you only read "utopia".
Utopia is good. But AGI "and" utopia sounds really, really.. i cant even @@alpha0xide9
56:50 I've got a really solid plan for AGI Alignment so I'll be looking to talk to #EmadMostaque very soon to get that Billion in funding. It's like nothing anyone else is considering, as far as I can tell, and it will be a whole lot of fun that brings the world together on an internal Alignment journey for humanity itself.
Wow
I doubt it. Usually people commenting in RUclips claiming they have a solution also have a case of Dunning Krueger. The real talents are rarely commenting on RUclips videos and claiming they have a solution to strangers. Instead, they are working in teams of other exceptionally gifted and talented people or also researching or in academia. If you’ve thought about a solution, they probably have already thought of it and discarded it. With the amount of actually intelligent people working on AI right now, there are probably dozens of ideas that are better than yours.
Well, Mostaque lied and stole his way to the top. So you can probably too.
@@douglasmyers-turnbull7136 is it just a conspiracy theory "out there" or is it something that is not known to the general public who really think Emad is a clever and creative guy???
@@douglasmyers-turnbull7136And here's Douglas, envious as ever
He is NOT the creator of stable diffusion, btw.
Extremely insightful; actually initially though Emad's face looked punchable 😂 but hes a great insightful guy
CEO stability AI Emad Mostaque is for humanity, transparently and open. CEO OpenAI Altman is for-profit closed Source.
I think the world need a national data source for the benefit that the people us what he al mention.
His quote what I saw in a other interview: "AI by the people, for the people".
Sorry, but he's actually complete trash who's been lying to everyone for years, even about things like his education. He's a massive fraud.
Painting it black and white is not a good idea. If Altman was all about profit, you'd think he would own a few shares of Open AI, no?
this aged well
He’s practically salivating watching society burn and we’re applauding him.
Hello? He is the one spearheading the open source side of things.
JESUS✝️REIGNS FOREVER MORE, HALLELUJAH!!!!!!! CELEBRATE AND GLORIFY JESUS EVERY DAY AND FOREVER MORE, HALLELUJAH!!!!!!! AMEN.
JESUS✝️IS THE ONLY HOPE FOR HUMANITY!!!!!!!
I LOVE YOU JESUS✝️
REPENT, FOR THE KINGDOM OF HEAVEN IS AT HAND.
Lord Jesus✝️have mercy on you🙏
This guy didn't create stable diffusion, he took the tech out of a university and named a company after it to make it seem like he did.
So, Language Models.. .
Natural Anti GABA - Gingko
Emad's moral compass is pointing correctly. Why is he not the CEO of OpenAI?
No, it's really, really not. He's been exposed as a fraud with no apparent moral compass. He lies to investors, reporters, and the public about pretty much everything, including that he's partnering with the WHO and even that he has a Master's degree (he doesn't). Elizabeth Holmes is probably more trustworthy, and her lies could have caused people to die.
Why is he not the CEO of some other company? What?
„The investor in Stable Diffusion“ would probably the more appropriate monicker - not „the creator“. And that shouldn’t discount him - you can’t create without investment
He started the company....
There’s a lot of coherence in what Emad says, but at the same time it seems a bit arrogant to imagine that the next years are so predictable. At this point in time, with the exponential growth of science and knowledge, it’s impossible to predict what’s going to happen even in the next month…
I mean he's a millionaire and his time is money and he got taken for a ride by autism hawksters/doomers. 3:00.
TLDR = To Lazy Don't Read. It is To Long Din't Read
Too Long; Didn't Read
@@alpha0xide9 Thank you
Lots of people chasing more.
43:25 ….Not many people were writing assembly 20 years ago. There were no libraries? I call BS ;) I was writing Haskell and Python 20 years ago. C++ had boost over 20 years ago. Emad is a fast talker and he talks BS sometimes ;)
Great interview. Emad should speak more clearly though.
Soon Emad Mostaque becomes the most richest man in the world and Stable Diffusion the most big opensource company in the world ...😮
Good God no. He can't even pay his employees on time. He's a massive fraud and habitual liar who's going to crash to the floor burning like Elizabeth Holmes did.
1:00:48 the more tech we have, the more devastating loneliness people will have
Why would anyone build an AI publishing business when your own personal AI on your device does all your personal news for you? All we need is AI reporters to feed JSON to the hive to propagate. Imagine, unbiased embodied AI journalists holding a flesh sack human politician to account. What a weird world.
It doesnt have to be good, it has to be used, and when it becomes used, they will trust it.
Hasn't he been going around lying! Lots of fakers in the world... another Elizabeth Holmes. Doing the lizzie!!😂
Our rotund and misshapen overlords. No beautiful future can come from people like this
Mishapen lol
Congrats, the most idiotic comment on here ❤
I think google will be the AI winner, they have the information and organization infrastructure.