David Sacks. On The All-in Podcast, he made a very major announcement. He said that artificial intelligence will blow up 1 million times in the next four years then, emphasizing that this has been all on the link and did not say it just for the hell of it. He further broke on three premises where AI is growing by huge multiples namely: the models, chips, and compute-the first model.
Advertisement
“AI algorithms advance by 3-4 times a year,” alleges Sacks. It was only a simple stepping stone, like a very basic chatbot; it has progressed into reasoning models, and now we are coming close to AI agents-that is, I would say, the next big leap. I would argue that we haven’t even scratched the surface. Then, there are also the hardware components. Chips are getting ever faster and more efficient, with entities like Nvidia finding ways to connect these chips together to give them a more powerful performance improvement. And lastly, compute power is shooting off the list of prices. Grok by Elon Musk started off with 100K GPUs, just as they went on to 300K, and next will be a million. OpenAI has embarked on a similar trajectory with its Stargate data center.
Taking all of those three things-10x improvements every other year-then you get a hundred times in each of those. Multiply them; boom-one suddenly sees a million. All of that, of course, will argue for cheaper AI, for smarter artificial intelligence, or of simply more artificial intelligence. The impact? Pretty much anything. But, as Sacks succinctly puts it, most people are oblivious to how fast this all really occurs since humans suck at understanding exponentially growing systems.
Time to get into the reactions because, as always, Twitter had thoughts. For some, it was a matter of taking Sacks i.e. @tradecryptoBtcX who just said, “In Sacks we trust.” “Then why Nvidia stock is falling,” says @1Entrepreneur, contemplating the apparent paradox that while demand has resounded for AI, Nvidia’s share declines. Then, of course, there’s @Munash_Tondoya, who said it la Sacks being a liar (with a smirk emoji, so we know they’re joking…probably).
Well, they weren’t the only replies to take a philosophical turn. @TrevorAVaughan pontificated on how little the human brain consumes in energy as compared to AI, suggesting that we still have a long way to go. And, of course, @kim_farry had a considerably longer saddish story about how AI was involved in diagnosing a visual deficit and much more reality than mere tech hoopla.
That won’t convince everybody, though. @jigarjianari replied-the models are not improving at 300% year on year, as in the case of GPT-2 to GPT-4. But @ChristosVK took a much darker tone, whining that “We are doomed” before imploring Sacks to do something about it.
Then came the wildcards. @MarthaJohn1679 admitted that she asked Gemini to explain exponential growth and got ‘squiggles’ she could not decipher. And @cheekykmk engaged mused about whether they are even talking to a real human being or a bot like Grok (which, by the way, actually replied back with a cheeky message confirming its bot status).
What’s the point, then? It doesn’t even matter whether you buy Sacks’ million-fold claim or think he has overhyped it; one thing is for sure: AI will be a part of life here onward. Not if it is going to change everything but just how fast and how much it is going to change it. And if Sacks is even half-right, it’s going to be a wild ride in the next few years.
Advertisement
Oh, and if you’re still not buying it, just consider that people once thought the idea of a smartphone was sci-fi. Now we’re arguing with chatbots on Twitter. So, million-fold? Probably not so far-fetched, after all.
