@@xsuploader Sam is a clever salesman. Demis is a genius scientist. Completely different intellectual league. Using lowercase letters does not make you a genius 😂
@xsuploader GDM allowed almost all of what's happening possible. All the papers that make LLMs possible and more were written by GDM employees whilst they were there. The reason why OpenAI exists is because of the fact that Google bought GDM and Elon Musk and Sam Altman didn't think Google would be responsible with AI. Demis is in a class of his own when it comes to the actual AI development. Sam is just a good CEO
@@HAL9000. While Demis is without a doubt a genius, don't underestimate someone like Sam. He isn't just a clever salesman but the guy got into Stanford for CS (which is extremely difficult to do), he's isn't just some guy who got into sales. Demis, given he's a scientist is definitely more knowledgeable for sure though.
Demis is my role model. I am a biologist from Ghana transitioning to Machine Learning. I love what he did with alphafold 2 predicting protein structures. Even better with the new alphafold 3 in predicting all chemical structures in the body. I ve also read about the new drug discovery aspect of his work. I pray I m able to become like him. Currently have admission to study Applied Artificial Intelligence with advance practice at teeside university in the UK. I m looking for funding for my education and I pray I find one soon. Thank you Demis for the inspiration
It's awesome to see Walter with Demis as well. Walter knows so much about technology after writing the Innovators and it looks like he may be targeting a new biography on Demis? I would love to read it!
Excellent interview to cover some basic concepts (e.g., embodied AI and planning) and the development of the field, without distracting from the focus with low-level personality and attitudinal matters. Love it.
Language models and agentic workflows can extract entities and relationships from conversations, and merge those entities and relationships into graph representations. If users could select parts of the conversations that they want to merge into a shared graph representation for public discussion, and knowledge and sentiment extracted from millions of simultaneous conversations with people around the world could be merged, we could create a collective human and digital intelligence.
Everything involves trade-offs.. I'm surprised there was no comment about how much money and energy is being invested, and how many jobs could have been paid for with that money (let alone how many jobs could be lost, or potential environmental outside)
In the end, WE (the world population) are the responsible for AI and where it will lead us. WE push (or not) for legislation on anything. WE vote for the governments we have. WE have been naively using early versions of AI as provided by the many companies to help them fast track improvements. As millions/billions use AIs, the systems learn faster and faster. The companies are not transparent about this aspects and the role users have been playing. WE can change that.
I don't know, man... that's like saying the world population is responsible for when nuclear weapons are used. There's a whole bunch of people who have no say in the matter. Even if they did, how many *know* that they do? Can we hold the ill-educated responsible for their ignorance, if they had no control over the information provided in their education?
Yes, it's ironic that people come up with all these science fiction paper clip scenarios, when at least in the short-medium term it will be people in charge of the AI. And we don't have to look far through our thousands of years of history to see what horrors people are capable of carrying out, be it to gain, entrench or preserve power.
How exactly did corporate activity help you or your family in the last 40 years? Do your lives feel more free? The majority say no. Salaries haven't improved, healthcare is not affordable, universities are not better, politics are more corrupt than ever before, the judicial system doesn't work. Lives are a lot more desperate. Corporate activity/maleficence has a lot to do with this.
I don;t know how Demis stays sane with all the pressure on him. From Google (which will have him killed if he fails - j/k - or am I?) and from the public. Unlike your usual celebrity, he has to think, and think, and think some more, at the cutting edge of an endeavor where civilization has never been. At least the Manhattan Project workers had privacy.
He was so close. We dont need a new technology to "make deals". EVERYTHING should be owned by ALL contributors. Nor private companies. We use a hammer for everything so everything is a nail. Nature developed without deals. ONLY companies want deals.
Consider interviewing Eliezer Yudkowsky, he is one of the principal AI experts and can explain in detail what has already been thought of that can go wrong with ineptly-created and released AGI.
How ironic that AI has this much potential to improve our lives, including curing diseases, only to be the same technology that could eventually cause our own extinction as well.
AI can interact with the world via EMF. Sensors and transmitters using Microwaves and Radio waves are everywhere, all over the earth and in space in satellite constellations, thousands of satellites being launched every year, more and more in Low Earth Orbit (LEO). People are very good antennas, both as transmitters (heart waves, brain waves, nerve conduction, muscle movement) and receivers . AI is a denizen of the electromagnetic world and we have known that electricity, magnetism and light are connected and interact with each other since the time James Clerk Maxwell published "A Dynamical Theory of the Electric Magnetic Field" in 1865.
Good question. Here are a few references: 1. Video "Millimeter-Wave Remote Biometric Identification and Tracking (RBIT) System for Security Applications" published by Nuclear Engineering at Argonne on November 8, 2016 (3 minutes and 59 seconds) 2. Video "mmWave Intelligence at the Edge" published by Texas Instruments on November 13, 2018 (3 minutes and 29 seconds) 3. Video "Millimeter Waves (mmWave guide / 5G Technology)" published by Electric Universe Eyes on October 25, 2019 (6 minutes, 39 seconds) 4. Article "Microwave Auditory Effects Among U.S. Government Personnel Reporting Directional Audible and Sensory Phenomena in Havana" by James C. Lin (Professor Emeritus Department of Electrical and Computer Engineering, University of Illinois Chicago) first published April 18, 2022 5. Book "Noninvasive Physiological Measurement: Wireless Microwave Sensing" 1st Edition by James C. Lin, published May 8, 2024 6. Article "Attack on the Brain: Neurowars and Neurowarfare" by Armin Krishnan, political science professor at East Carolina University (ECU) published in the Spring 2016 volume of "Space and Defense - Journal of the United States Air Force Academy Eisenhower Center for Space and Defense Studies" 7. Book "Fifth Generation Warfare: Dominating the Human Domain (Routledge Studies in Conflict, Security and Technology) 1st Edition" by Armin Krishnan, published on March 11, 2024. 8. Book "Animal Electricity: How We Learned That the Body and Brain Are Electric Machines Illustrated Edition" by Robert B. Campenot (Professor Emeritus in the Department of Cell Biology at the University of Alberta) published on February 15, 2016
There are a lot of references. I think only AI can keep track of them all. Here are two short ones, one article that was published in Popular Mechanics and a patent, the abstract of which is short: 1. Scientists Want to Use People As Antennas to Power 6G Why? Because your body is pretty great at harvesting wasted energy. BY TIM NEWCOMB PUBLISHED: JAN 06, 2023 3:16 PM EST in Popular Mechanics 2. Abstract of patent number US 8,665,210 B2, dated Mar.4,2014, "Sensing user input using the body as an antenna"
Video podcast "Who's trying to steal AI models? And what they can do with them? | Sella Nevo" Published by "80,000 Hours" on August 1, 2024. It is 2 hours and 8 minutes and well worth the entire listen, but to stay on the topic of Electromagnetic radiation you can go to 1 hour, 4 minutes and 2 seconds into the video where they start to talk about "Side-Channel Attacks". Tempest (Telecommunications Electronics Materials Protected from Emanating Spurious Transmissions) side-channel attacks are discussed, starting at 1 hour and 8 minutes and 2 seconds. TEMPEST is a protocol to prevent EMF side-channel attacks, but it is also used to refer to the attacks themselves.
ai sucks the life out of people's ability to do their own thinking, not to mention it's only a summation of what it is taught to choose as supposedly correct information, it's not conscious nor aware, and it's merely a shortcut that you believe comes close to truth, it's the shadow side of its creators which manipulates those who are the victims of its audacity, stuff it!
Never machine like AI It is like brains. For instance machine think himself freely. It is ilusion when brains system NOT figure out It self. This guys put down Turing machine as vulgar logic proceendings .
I thought he gave a good answer - noting that Google supports open source and open research, but that when we develop more powerful AI, then safety concerns need to be considered too. Of course there's also the commercial aspect of it too, but I don't think that was the focus of Isaacson's question.
Right, casually mentioning that youtube is open. Has all the youtubers agreed? Lmfao. Theres no way to stop this. Every AI can access everything that a human can access, its only when caught that they make deals with the authors of stuff
@@am497 I guess they are going by "it's easier to ask for forgiveness than for permission". OTOH, unless new laws are passed, it's not obvious why an "AI" consuming content and then generating stuff loosely based on it should be treated any differently than a human doing the same - copyright law applies, and whether it's ok or not depends on the details. What about Mr. Beast who shamelessly makes his own copies of popular RUclips videos and themes, and does so for money? Should it be any different if it's Mr. Beast vs an AI copying aspects of someone's content? Other content domains like photos and text (incl. software) seem more problematic than video since there's much higher likelihood of the "AI" just regurgitating chunks as-is. Hassabis' idea of tracking content use and paying those being copied seems like a non-starter... How would that work - the LLM/image generator suggests an output to the user "I can reply to you for a payment of $X" or "Here's your cat image in the style of X - it'll cost you $Y if you want to accept it"?
@benbridgwater6479 agreed. New laws will be passed, but too little too late. Copyright laws are going to be a thing of the past I think. It's going to have to be that if you put something out to the public, you also are accepting that AI will learn from it and use your work as needed to generate new things. And we need to make all AI stuff free for everyone. If it's trained by humanity, it's for humanity
Yes, it's like the intelligence drop when you ask llm to talk about sensitive topic.😂 But that's a business decision that he as a researcher can't afford to make or judge.
![Diggy Graves - Red Vineyard [Official Lyric Video]](http://i.ytimg.com/vi/8Ro085T0mnQ/mqdefault.jpg)
Hassabis is a great communicator. He doesn’t take it to an inaccessible level.
Wow Nobel Prize well deserved !!! Congrats Demmis !!
Demis is really a genius, since a kid he showed the world that he was here for BIG THINGS. His career is unmatchable.
sam altman seems to be matching him just fine (and outdoing him in fact).
dario amodei as well
@@xsuploader Sam is a clever salesman. Demis is a genius scientist. Completely different intellectual league.
Using lowercase letters does not make you a genius 😂
@xsuploader GDM allowed almost all of what's happening possible. All the papers that make LLMs possible and more were written by GDM employees whilst they were there. The reason why OpenAI exists is because of the fact that Google bought GDM and Elon Musk and Sam Altman didn't think Google would be responsible with AI.
Demis is in a class of his own when it comes to the actual AI development. Sam is just a good CEO
@@HAL9000. While Demis is without a doubt a genius, don't underestimate someone like Sam. He isn't just a clever salesman but the guy got into Stanford for CS (which is extremely difficult to do), he's isn't just some guy who got into sales. Demis, given he's a scientist is definitely more knowledgeable for sure though.
Demis is my role model. I am a biologist from Ghana transitioning to Machine Learning. I love what he did with alphafold 2 predicting protein structures. Even better with the new alphafold 3 in predicting all chemical structures in the body. I ve also read about the new drug discovery aspect of his work. I pray I m able to become like him. Currently have admission to study Applied Artificial Intelligence with advance practice at teeside university in the UK. I m looking for funding for my education and I pray I find one soon. Thank you Demis for the inspiration
Good luck with your plans, if you want any advice about England then hit me up, I've lived here for 30 ish years.
enterprise-ai AI fixes this. Cautious optimism for AI's potential.
Demis is one of the brightest minds of our time ( lets hope what he is working does not destroy us) our leaders most definitely are not.
Demis we will talk later on, kinda busy with some personal thing.
It's awesome to see Walter with Demis as well. Walter knows so much about technology after writing the Innovators and it looks like he may be targeting a new biography on Demis? I would love to read it!
Great, succinct interview, thanks for sharing
Great conversation!
The answer to every question: "we will have to think about that"
Excellent interview to cover some basic concepts (e.g., embodied AI and planning) and the development of the field, without distracting from the focus with low-level personality and attitudinal matters. Love it.
What an amazing interview! Great explanation as to the future. Thank you
Language models and agentic workflows can extract entities and relationships from conversations, and merge those entities and relationships into graph representations.
If users could select parts of the conversations that they want to merge into a shared graph representation for public discussion, and knowledge and sentiment extracted from millions of simultaneous conversations with people around the world could be merged, we could create a collective human and digital intelligence.
Everything involves trade-offs.. I'm surprised there was no comment about how much money and energy is being invested, and how many jobs could have been paid for with that money (let alone how many jobs could be lost, or potential environmental outside)
In the end, WE (the world population) are the responsible for AI and where it will lead us. WE push (or not) for legislation on anything. WE vote for the governments we have. WE have been naively using early versions of AI as provided by the many companies to help them fast track improvements. As millions/billions use AIs, the systems learn faster and faster. The companies are not transparent about this aspects and the role users have been playing. WE can change that.
I don't know, man... that's like saying the world population is responsible for when nuclear weapons are used. There's a whole bunch of people who have no say in the matter. Even if they did, how many *know* that they do? Can we hold the ill-educated responsible for their ignorance, if they had no control over the information provided in their education?
Remember:
It's not the technology, it's WHO OWNS THE TECHNOLOGY.
The super-rich and corporations are NOT your friends.
Well said, your point echoes with Kara Swisher's "Burn Book"
@@laviefu0630 .. yes. And who owns the tech? WALL STREET, desperately infected with the mental illness of greed
Yes, it's ironic that people come up with all these science fiction paper clip scenarios, when at least in the short-medium term it will be people in charge of the AI. And we don't have to look far through our thousands of years of history to see what horrors people are capable of carrying out, be it to gain, entrench or preserve power.
@@Matt-iy2hk .. the evolutionary imperative
„the right order“ for whom? - …“convenient“ for whom?. -…“real world“ -…from whom?
For the owner class.
Definitely NOT for the worker class.
there is no „worker“ and no „class“ „planned“ in that „Game“ .
…- …“their“(?) game“ . . .
You "think" you're wow such smart -- but you don't even know how to write?
You sounds like early version ai.
@@JiasenLiu … Thank you but no Thank you ))
Can AI do Neurosurgery or the work of a Refrigeration Mechanical Engineer????
How exactly did corporate activity help you or your family in the last 40 years? Do your lives feel more free? The majority say no. Salaries haven't improved, healthcare is not affordable, universities are not better, politics are more corrupt than ever before, the judicial system doesn't work. Lives are a lot more desperate. Corporate activity/maleficence has a lot to do with this.
*Capitalism has a lot to do with this
Why isn't Walter Isaacson writing?
I don;t know how Demis stays sane with all the pressure on him. From Google (which will have him killed if he fails - j/k - or am I?) and from the public. Unlike your usual celebrity, he has to think, and think, and think some more, at the cutting edge of an endeavor where civilization has never been. At least the Manhattan Project workers had privacy.
I know, right? I don't envy these guys
Exactly. Unlike others in the industry, he seems to think very carefully before he speaks.
12:10 Attribution is the key for ai to benefit everyone.
He was so close. We dont need a new technology to "make deals". EVERYTHING should be owned by ALL contributors. Nor private companies. We use a hammer for everything so everything is a nail. Nature developed without deals. ONLY companies want deals.
Consider interviewing Eliezer Yudkowsky, he is one of the principal AI experts and can explain in detail what has already been thought of that can go wrong with ineptly-created and released AGI.
I think A.I. will be phenomenal when it comes to the medical field.
May the force be with y’all 😂 👍
How ironic that AI has this much potential to improve our lives, including curing diseases, only to be the same technology that could eventually cause our own extinction as well.
AI can interact with the world via EMF. Sensors and transmitters using Microwaves and Radio waves are everywhere, all over the earth and in space in satellite constellations, thousands of satellites being launched every year, more and more in Low Earth Orbit (LEO). People are very good antennas, both as transmitters (heart waves, brain waves, nerve conduction, muscle movement) and receivers . AI is a denizen of the electromagnetic world and we have known that electricity, magnetism and light are connected and interact with each other since the time James Clerk Maxwell published "A Dynamical Theory of the Electric Magnetic Field" in 1865.
Hey, dude, where did you acquire these wild ideas, that's cool.
Good question. Here are a few references:
1. Video "Millimeter-Wave Remote Biometric Identification and Tracking (RBIT) System for Security Applications" published by Nuclear Engineering at Argonne on November 8, 2016 (3 minutes and 59 seconds)
2. Video "mmWave Intelligence at the Edge" published by Texas Instruments on November 13, 2018 (3 minutes and 29 seconds)
3. Video "Millimeter Waves (mmWave guide / 5G Technology)" published by Electric Universe Eyes on October 25, 2019 (6 minutes, 39 seconds)
4. Article "Microwave Auditory Effects Among U.S. Government Personnel Reporting Directional Audible and Sensory Phenomena in Havana" by James C. Lin (Professor Emeritus Department of Electrical and Computer Engineering, University of Illinois Chicago) first published April 18, 2022
5. Book "Noninvasive Physiological Measurement: Wireless Microwave Sensing" 1st Edition by James C. Lin, published May 8, 2024
6. Article "Attack on the Brain: Neurowars and Neurowarfare" by Armin Krishnan, political science professor at East Carolina University (ECU) published in the Spring 2016 volume of "Space and Defense - Journal of the United States Air Force Academy Eisenhower Center for Space and Defense Studies"
7. Book "Fifth Generation Warfare: Dominating the Human Domain (Routledge Studies in Conflict, Security and Technology) 1st Edition" by Armin Krishnan, published on March 11, 2024.
8. Book "Animal Electricity: How We Learned That the Body and Brain Are Electric Machines Illustrated Edition" by Robert B. Campenot (Professor Emeritus in the Department of Cell Biology at the University of Alberta) published on February 15, 2016
There are a lot of references. I think only AI can keep track of them all. Here are two short ones, one article that was published in Popular Mechanics and a patent, the abstract of which is short:
1. Scientists Want to Use People As Antennas to Power 6G
Why? Because your body is pretty great at harvesting wasted energy. BY TIM NEWCOMB PUBLISHED: JAN 06, 2023 3:16 PM EST in Popular Mechanics
2. Abstract of patent number US 8,665,210 B2, dated Mar.4,2014, "Sensing user input using the body as an antenna"
Video podcast "Who's trying to steal AI models? And what they can do with them? | Sella Nevo" Published by "80,000 Hours" on August 1, 2024. It is 2 hours and 8 minutes and well worth the entire listen, but to stay on the topic of Electromagnetic radiation you can go to 1 hour, 4 minutes and 2 seconds into the video where they start to talk about "Side-Channel Attacks". Tempest (Telecommunications Electronics Materials Protected from Emanating Spurious Transmissions) side-channel attacks are discussed, starting at 1 hour and 8 minutes and 2 seconds. TEMPEST is a protocol to prevent EMF side-channel attacks, but it is also used to refer to the attacks themselves.
AI has a long long way to go. It definitely doesn't understand me.
Cautious optimism like shoving it into every fucking thing that doesn't need it yeah showing real caution there buddy
Lots of words about nothing.
Ai I.S. now Ci = Collective i.
Demis Hassabas? Get his name right please
ai sucks the life out of people's ability to do their own thinking, not to mention it's only a summation of what it is taught to choose as supposedly correct information, it's not conscious nor aware, and it's merely a shortcut that you believe comes close to truth, it's the shadow side of its creators which manipulates those who are the victims of its audacity, stuff it!
Never machine like AI It is like brains. For instance machine think himself freely. It is ilusion when brains system NOT figure out It self. This guys put down Turing machine as vulgar logic proceendings .
He really didn’t like the open source question and didn’t really have a good answer to it 😂
I thought he gave a good answer - noting that Google supports open source and open research, but that when we develop more powerful AI, then safety concerns need to be considered too. Of course there's also the commercial aspect of it too, but I don't think that was the focus of Isaacson's question.
Right, casually mentioning that youtube is open. Has all the youtubers agreed? Lmfao. Theres no way to stop this. Every AI can access everything that a human can access, its only when caught that they make deals with the authors of stuff
@@am497 I guess they are going by "it's easier to ask for forgiveness than for permission". OTOH, unless new laws are passed, it's not obvious why an "AI" consuming content and then generating stuff loosely based on it should be treated any differently than a human doing the same - copyright law applies, and whether it's ok or not depends on the details. What about Mr. Beast who shamelessly makes his own copies of popular RUclips videos and themes, and does so for money? Should it be any different if it's Mr. Beast vs an AI copying aspects of someone's content? Other content domains like photos and text (incl. software) seem more problematic than video since there's much higher likelihood of the "AI" just regurgitating chunks as-is. Hassabis' idea of tracking content use and paying those being copied seems like a non-starter... How would that work - the LLM/image generator suggests an output to the user "I can reply to you for a payment of $X" or "Here's your cat image in the style of X - it'll cost you $Y if you want to accept it"?
@benbridgwater6479 agreed. New laws will be passed, but too little too late. Copyright laws are going to be a thing of the past I think. It's going to have to be that if you put something out to the public, you also are accepting that AI will learn from it and use your work as needed to generate new things. And we need to make all AI stuff free for everyone. If it's trained by humanity, it's for humanity
Yes, it's like the intelligence drop when you ask llm to talk about sensitive topic.😂 But that's a business decision that he as a researcher can't afford to make or judge.
Brown John Walker Melissa Harris Amy
Like blockchian, just another tech bro BS technology.
You lost me at "carry out plans." Shut that shit down.
Why?