I asked Ron Howard why AI is always the bad guy in sci-fi movies, and he said "because it's the easy way to tell a story. If you want something different, write it." So I did:
One of the saddest things I've learned recently: students deciding against getting a PhD because they can't take the political correctness of academia any more.
Career paths in AI:
Good at math: learning theory
Good at GPU programming: deep learning
Good at data wrangling: data science
None of the above: AI ethics
Greatest textbooks of all time (1/2):
Feynman, "The Feynman Lectures on Physics"
Pauling, "General Chemistry"
Samuelson, "Economics"
Russell, "A History of Western Philosophy"
Knuth, "The Art of Computer Programming"
Step 1: Sundar Pichai is terrified Gemini will blow up, so he gives the responsible AI group extensive power to shape it.
Step 2: The responsible AI group goes to town making Gemini woke.
Step 3: Woke Gemini causes the blowup Sundar was terrified of.
A group at IBM developed a link-based search engine before Google. IBM refused to market it, saying it would never make money. A member of that group (Prabhakar Raghavan) is now Google's head of search and ads, responsible for more revenue per year than IBM's entire market cap.
The irony of Nvidia is that its whole game plan was to build application-specific chips, but it's now worth $2 trillion because its chips turned out to be general-purpose.
The psychology of AI alarmists:
Elon Musk: Savior complex. Needs something to save the world from.
Geoff Hinton: Ultra-leftist, world-class eccentric.
Yoshua Bengio: Hopelessly naive idealist.
Stuart Russell: His only impactful application ever was to nuclear test monitoring.…
What they're thinking right now:
Sundar Pichai: "Damn, I knew releasing a chatbot was a bad idea."
Satya Nadella: "Wow, we didn't just make Google dance, we made them fall on their face."
Sam Altman: "Yay, Gemini almost makes ChatGPT look like it's not woke."
Mark Zuckerberg:…
The Real AI 100
@TIME
's list of the supposedly 100 most influential people in AI is such a mess that as a public service I've taken the time to compile a more reasonable one:
Pieter Abbeel
@pabbeel
Ajay Agrawal
@professor_ajay
Sam Altman
@sama
Marc Andreessen
@pmarca
Susan…
Jensen Huang told me in 2016 that when Nvidiers first noticed AI researchers were using GPUs they were flabbergasted. The speed and force with which they then pivoted to AI is a case study for the ages.
2000s: Neural networks = Deep learning.
2010s: Machine learning = Deep learning.
2020s: Artificial intelligence = Deep learning.
2030s: Computer science = Deep learning.
2040s: Science = Deep learning.
2050s: The universe = Deep learning.
The woke method:
1. Willfully misunderstand and misrepresent what your opponents say.
2. Hammer on it until your misrepresentation prevails.
3. If that doesn't work, kill discussion by insulting everyone until they don't want to be there.
A Silicon Valley insider (Reid Hoffman) told me an Inflection co-founder (Reid Hoffman) negotiated a deal with a Microsoft board member (Reid Hoffman), advised by a Greylock partner (Reid Hoffman).
The ROI (return on intellect) of science is lowest in physics (massive intellect, no return) and highest in deep learning (no intellect, massive return).
All mathematics is produced by physical objects (brains, computers). Therefore the laws of physics determine what theorems are possible, and mathematics has no existence independent of our universe.
Interesting how in all these domains AI is asymptoting at roughly human performance - where's the AI zooming past us to superintelligence that Kurzweil etc. predicted/feared?