A while ago Son said ASI would arrive in 2043, now 2035, he also defines it as an AI with 10,000x the intelligence of humans - which is probably a decent benchmark. But he also says this AI will use 400Gw of energy and 20 Billion GPU's - today we're seeing new AI accelerator chips that use 20x less power with new AI training models that also use 1,000x less compute (and less energy), while Nvidia have said their next generation GPU (by 2030) will be 1,000,000x more powerful than their current Blackwell line so when it comes to the chips needed and the power requirements I think he's still thinking too "today" and too linearly (ironically). He also says we need to regulate ASI but only AI could regulate this kind of super powerful fast evolving AI system, and also says criminals won't have access to these models because they would need HUGE amounts of capital to create them and no criminal organisations have the money needed - however he ignores the fact that states like China (and other governments to be fair) DO have the money to create these models and then use them to hack other nations etc. Overall I was surprised by his comments and linear thinking .... and he's in charge of a HUGE multi-billion dollar fund that steers the future of this technology so I'm not sure why he is so far off ... ?!
good guy, honest af, he is actually genuine cause he's talking about just only one superrun of /ai and then all major problems can be done. 9 trillion tho
Like him or not, he is a legend, totally an unusual human.
I somehow like to listen to him. Down to earth and feels relatable.
A self-made man, Always optimistic about new tech. Last minute is excellent.
13 minutes with Masa is not enough. He was just beginning to cook with that $9T investment comment, but the interviewer cut him off.
He is a legend. He is not talking big. It is possible when it becomes true.
the last minute is priceless.
One and Only, Masa! What a legend.
His way of thinking big & putting into perspectives (I.e. 5% of gdp) is amenable & admirable
i like him, he's always optimistic
A while ago Son said ASI would arrive in 2043, now 2035, he also defines it as an AI with 10,000x the intelligence of humans - which is probably a decent benchmark. But he also says this AI will use 400Gw of energy and 20 Billion GPU's - today we're seeing new AI accelerator chips that use 20x less power with new AI training models that also use 1,000x less compute (and less energy), while Nvidia have said their next generation GPU (by 2030) will be 1,000,000x more powerful than their current Blackwell line so when it comes to the chips needed and the power requirements I think he's still thinking too "today" and too linearly (ironically). He also says we need to regulate ASI but only AI could regulate this kind of super powerful fast evolving AI system, and also says criminals won't have access to these models because they would need HUGE amounts of capital to create them and no criminal organisations have the money needed - however he ignores the fact that states like China (and other governments to be fair) DO have the money to create these models and then use them to hack other nations etc. Overall I was surprised by his comments and linear thinking .... and he's in charge of a HUGE multi-billion dollar fund that steers the future of this technology so I'm not sure why he is so far off ... ?!
good take!
Absolutely, Mr. Masayoshi Son has a clear and correct vision of AI. AI is our hope because it can help people and make us happier🙏
Well, he is a character
The guy doing the interview is annoying as hell. Just ask the question and leave the antics alone.
لغة الجسد تخبر عن روعة التفاعل من الشخص المتحدث أولا .. هو المذيع ؟
ما خبراته السابقة يا ترى ؟
هل يمكن الإفادة ..
أعجبني جدا ،،
A bad guy may not, a bad gov can
I think his projections are somewhat conservative. I think ASI will be somewhat before '35 at much less than $9T investment.
Whoever in the world can build an antidote to the bad A.I technology will be a self made billionaire.
As he said , world will change a lot in next 10 years . Just because of AI , the change will be disruptive.
Who will be able to afford these humanoid robots if they become the main employee in the work force making humans incomeless.
The guyz lost everything in uber.
Musk & Nvidia will outcompete openAI and ARM
Nvidia - yes, Musk - no
Irritating interviewer, bad choice!
good guy, honest af, he is actually genuine cause he's talking about just only one superrun of /ai and then all major problems can be done. 9 trillion tho