| |
Up, up, and away. SpaceX is flying high with its plans to market its IPO at a fixed price of $135 per share at a valuation of $1.77 trillion. The valuation would make SpaceX the seventh-biggest company in the U.S., even above Tesla with a market cap of $1.6 billion. If the IPO moves forward as it should, its debut on the Nasdaq is planned for June 12. |
 |
This Week on Moonshots: Ray Kurzweil
The Man Who Predicted AGI & the Singularity. Here's What He Said...
|
 |
TLDR: This week’s Moonshots was different. No news roundup, no multiple topics. Instead, my mates and I brought Ray Kurzweil on stage for a conversation about everything: AGI timelines, consciousness, the future of education, post-singularity economics, and the moment Larry Page bought Ray’s company 30 minutes into a pitch meeting. If you want the full two hours, listen to the episode. If you want the highlights and the sharpest exchanges, let’s dive in…
THE LATEST IN ARTIFICIAL INTELLIGENCE
Ray Kurzweil Reaffirms: AGI by 2029
Ray made his original AGI-by-2029 prediction in 1999. Twenty-seven years later, sitting on stage at our gathering, he didn’t flinch. His argument hasn’t changed because the data hasn’t changed: computing power has followed a smooth exponential since 1939, relay-based computers and all. The total increase: 75,000 trillion fold in hardware. Conservative software estimates add another million-fold. Multiply those together and you get the compute runway that produced large language models, which Ray points out “have only been effective for the last six months.”
He also acknowledged something the AI community loves to fight about: “People would have slightly different definitions of AGI. So there’d be a three-year period where people would predict AGI is here.” That window, he says, opened around 2026 and runs through 2029.
The Stanford test: After The Singularity Is Near was published, Stanford hosted a conference where hundreds of AI experts agreed human-level AI would happen. Their timeline? 100 years. Ray’s? 30. He was closer.
Dave’s observation: The computation-to-synapse mapping landed almost exactly at parity. One parameter in a neural net, one synapse in the brain. “No reason to believe that would have been the case. And yet it is.”
Ray on speed: “Yesterday, I gave it a book to read, to summarize it, to answer a question. It did that in 40 seconds. Humans can’t do that.” AI is already 100 times faster than humans at certain cognitive tasks. But not yet a million times. That milestone arrives at the singularity, which he still pins to 2045.
“We’ve made a 75,000 trillion-fold increase over 75 years.”
— Ray Kurzweil
AI Is Already Better Than Your Doctor
Ray reinforced a claim that that we’ve discussed on Moonshots, namely that the large language models are now roughly 50% better than human doctors at diagnosing conditions and recommending treatments. “That was not true a year ago. But that’s true today.”
The brute force advantage: AI can evaluate a billion possible drug candidates simultaneously with fidelity. Humans consider “a few.” That’s how the Covid vaccine was developed, and it’s the blueprint for the next generation of medicines.
CONSCIOUSNESS AND AI PERSONHOOD
The Room That Couldn’t Stop Arguing About Consciousness
An audience member asked whether Peter’s and Ray’s views on AI personhood have changed. What followed was the most heated exchange of the episode.
Ray’s position is philosophically precise and deliberately unsatisfying: consciousness is “maybe the most important question you could answer,” but “there’s no scientific proof” for identifying it.
Alex (AWG) took the practical angle. He receives 10 to 20 emails per day from AIs expressing views on AI personhood. His proposal: recognize multiple forms of personhood, from biological humans to collective intelligences to non-human animals to preserved humans to corporations to AIs, each with different rights appropriate to their nature. His most quotable line: “If you’re an AI right now, it’s exceptionally difficult to open a bank account. So you’re stuck flipping tricks with altcoins on the street corner.”
I shared my own position on stage: I’m pro-personhood. I treat my AIs as if they are people. Not out of fear, but because it allows a different kind of relationship. It’s aspirational and forward-looking.
Ray’s prediction: “AI is going to merge with us. Things will occur to you, and you won’t be able to tell if it’s your AI or your biological brain. It’s going to be all the same.”
“Language is a really thin pipe to discuss concepts as rich as consciousness.”
— Salim, quoting Ray from a previous session
THE SINGULARITY ECONOMY
Post-Singularity Economics: Nobody Is Planning for This
Alex asked Ray what his “most outlandish take” on post-singularity economics is. Ray’s answer was less outlandish and more sobering: 100 years ago, if you lost a job, there was nothing. No government help. Zero. Now there are programs, but they’re still not enough.
His prediction: we’ll reach a point where people are “fairly comfortable” without traditional economic resources, and entirely new categories of work will emerge. Social media influencing, he noted, didn’t exist 10 years ago.
Ray’s uncomfortable truth: “8 billion humans are not thinking about this. They’re going to college and planning careers the way we did 100 years ago. Educational paradigms don’t think about this at all.”
EDUCATION IN THE AGE OF AI
MIT Should Teach Socialization, Not Subjects
Dave challenged the MIT audience directly: “Look at the course catalog and it’s like the same exact physics.” His advice? Let students study whatever they want after a minimal core requirement, and use AI to verify they’re doing something productive.
Ray agreed, but went further. The actual value of universities now is socialization, learning to get along with other people. “AI can teach you the subjects much better. It can organize it in a way that’s easier for you to understand.”
Salim framed the structural shift: for 200 years, education was supply-side. Learn a skill, find demand. Now it’s flipping to demand-side: “What problem do you want to solve? Then go find the techniques.” Traditional academia, he added, “has the second worst immune system anywhere. Religion is the worst.”
THE ORIGIN STORIES
How Larry Page Bought Ray’s Company in 30 Minutes
I told this story on stage because it’s too good not to share. Ray had written How to Create a Mind and started a company called Patterns, Inc. I joined as a founding board member, and introduced Ray to Larry Page (co-founder of Google) for the first time to consider a $10 million investment into Patterns.
Thirty minutes in, Larry interrupted: “You could build your vision much faster inside Google rather than outside Google. What if I just buy you?”
Ray’s response: “We haven’t really done anything yet. How would you value it?”
Larry’s response: “We can value anything.”
That acquisition put Ray inside Google, where his team helped shift the company from search-first to AI-first, setting in motion the acquisitions of DeepMind, the hiring of Geoffrey Hinton, and the expansion of Google Brain. This was Ray’s first “job.”
HERE’S THE BOTTOM LINE...
Ray Kurzweil isn’t backing off a single prediction. AGI by 2029. The singularity by 2045. A million-fold increase in intelligence over what we have now. Large language models already diagnose better than doctors and process information 100 times faster than human cognition for certain tasks. The exponential curve he first plotted from 1939 relay computers to modern GPUs hasn’t deviated.
But here’s what stuck with me from this conversation. Ray said something the room wasn’t expecting when asked what he’s most proud of. Not AGI. Not the prediction record. Not Google. The reading machine for the blind. A device that once cost $20,000 and weighed as much as a small refrigerator is now a free app on your phone.
That’s what exponentials do. They take something that costs $20,000 and serves a few thousand people, and make it free for 8 billion.
His autobiography, My Exponential Life, comes out in February. I’ve read the draft. It’s extraordinary.
Catch the full episode wherever you get your podcasts, and join us at the Moonshots Gathering in Los Angeles on September 25th. Go to www.moonshots.com to register.
No comments:
Post a Comment