A lot this week - so in prioritized order:
- Cognition Labs launches Devin - the first AI software engineer. The demo is incredible. Impatiently awaiting access!
- LLM capabilities are growing far faster than Moore's Law!
- A paper on AI and creativity/art. "Our research shows that over time, text-to-image AI significantly enhances human creative productivity by 25% and increases the value as measured by the likelihood of receiving a favorite per view by 50%. While peak artwork Content Novelty, defined as focal subject matter and relations, increases over time, average Content Novelty declines"
- Marc Andreessen and Tyler Cowen talk AI
- Another favorite economist - Tim Hartford - on what the birth of the spreadsheet teaches us about generative AI.
- A Hacker's Guide to Language Models. First 30 minutes is very accessible and the rest is fun as well.
- A video from a brand x AI conference last year that I somehow missed. "LLMs are bad at things we expect computers to be good at, they are good at what we expect computers to be good at." An excellent insight! I'll be going to this years conference.
- LLMs automating social science. "Starting from a social science “scenario,” an LLM is queried to propose outcomes of interest and the relevant agents. For each outcome proposed, potential exogenous causes are again elicited from the LLM. For example, a scenario of interest might be “two people bargaining over a mug.” The LLM may propose “whether a sale occurs” as an outcome of interest, with a buyer and a seller as the relevant agents. The LLM might hypothesize that the buyer’s willingness to pay may affect the outcome…"
- McKinsey on Generative AI and the future of New York - "This growth will be partially offset by 600,000 positions that could be affected in office support, customer service, and food service and production, precipitated by accelerated automation"
- A great read on how AI is and will impact news from the editorial director of AI initiatives at The New York Times
- Ethan Mollick launches a site of AI resources (ie: idea generation prompt) and writes on how much prompting matters (TLDR; a lot).
- A really nicely balanced New Yorker write-up on AI Doomsayers
- I've been reading a lot about the history of early computing. Here is an essay from 1955 from Jon von Neumann - "Can We Survive Technology"