AI

Artificial intelligence as a field and an investing theme. The wiki uses this page as the umbrella tag for the field; the current generative wave has its own page at Generative AI, and the models layer is covered under Foundation Models and Machine Learning.

How it shows up

• Who captures value in each wave. Elad Gil’s thread on tech waves holds that the last AI wave (roughly 2014–2019) sent almost all value to incumbents (Google, Netflix recommendations, Facebook, TikTok), despite many machine-learning startups. He offers three candidate reasons: the technology was not 10X better than incumbent products and distribution, large data sets were the moat, and many startups chose extra-hard markets such as healthcare. He argues the generative wave looks different: much stronger technology, new infrastructure companies, and clearer application use cases. See (1-N) Every Tech Wave Results… (tweet) and Startup vs Incumbent.
• Hype and skepticism. Anna Khan asked in October 2022 whether generative AI was “the next web3/crypto craze,” questioning the valuations of companies built on open-source technology (🚨 1- I’m Still Learning… (tweet)). On the Contrary GP episode, AI is called “the clear hype cycle,” which the speaker treats as typically the worst time to invest (Contrary GP on Talent and Research-Driven Investing; see also Hype Cycles).
• Pace and cost. Saved posts track the economics and trendlines: METR time-horizon research (More Great Metr Research In… (tweet)), a Dwarkesh Patel post on why compute could become 10x or more expensive if a leading lab reaches $1T in revenue (New blog post on what… (tweet)), and Dan Primack’s point that sober conversation is hard when so much money is at stake (It’s Hard To Have A Sober… (tweet)).
• Quirks of use. Ai Researchers Are Now Injecting… (tweet) notes authors hiding prompts in papers because some reviewers use ChatGPT; Everyone is Reading Lonesome Dove records Claude and ChatGPT both recommending the same book.
• Reading. Deep Learning (Goodfellow) and The Scaling Era sit in the reading queue as the field’s textbook and oral history.

Where this appears

Tweets

• (1-N) Every Tech Wave Results… (tweet)
• 🚨 1- I’m Still Learning… (tweet)
• More Great Metr Research In… (tweet)
• New blog post on what… (tweet)
• It’s Hard To Have A Sober… (tweet)
• Ai Researchers Are Now Injecting… (tweet)
• Fascinatingly, the machine-learning techniques used… (tweet)

Podcasts and long-reads

• Contrary GP on Talent and Research-Driven Investing
• Everyone is Reading Lonesome Dove

Books to read

• Deep Learning (Goodfellow)
• The Scaling Era