I have helped run an AI-based entrepreneurship program for years, written papers on the econ of AI, and follow the field quite closely. Nonetheless, I am *shocked* by how good OpenAI's new chat (chat.openai.com/chat) is. E.g., you can no longer give take-home exams/homework.

ChatGPT is the best bullshitter there's ever been. Despite the hype, it won't replace search and isn't a good learning tool. But it's still exciting! Sayash Kapoor and I identify three specific areas where ChatGPT is very useful despite its limitations. aisnakeoil.substack.com/p/chatgpt-is-a…

Everyone is worried about concentration in markets. Is it actually increasing? As usual for these big macro trends, it depends. A 🧵 on the recent literature

Three years in the making - our big review/position piece on the capabilities of large language models (LLMs) from the cognitive science perspective. Thread below! 1/ arxiv.org/abs/2301.06627

🚨🚨🚨Announcing the 2023 SUMMER INSTITUTES IN COMPUTATIONAL SOCIAL SCIENCE 🚨🚨🚨 This program will run in 28 locations around the globe this year! Learn more (and apply) at: sicss.io/locations

In the age of rapid technological advancement, I often ask myself, what value can social scientists bring to the table? Technology evolves faster than us. The problems that will haunt us are more likely to be human problems.

Here it is, everyone, the first good model of which industries & jobs are most at risk of disruption from the sudden boom in generative AI, using an established method that looks at job tasks. If your job or industry is on these lists, you should be paying close attention to AI.

Another day, another opinion essay about ChatGPT in the The New York Times. This time, Noam Chomsky and colleagues weigh in on the shortcomings of language models. Unfortunately, this is not the nuanced discussion one could have hoped for. 🧵 1/ nytimes.com/2023/03/08/opi…

Large language models change everything for linguistics, starting with Chomsky. Featuring: LLMs as scientific theories, response to prior takes, "why" questions in language, acquisition... and how the field should have seen this coming. Paper is here: lingbuzz.net/lingbuzz/007180

How do we compare the scale of language learning input for large language models vs. humans? I've been trying to come to grips with recent progress in AI. Let me explain these two illustrations I made to help. 🧵

People are testing large language models (LLMs) on their "cognitive" abilities - theory of mind, causality, syllogistic reasoning, etc. Many (most?) of these evaluations are deeply flawed. To evaluate LLMs effectively, we need some principles from experimental psychology.🧵

Reminder that, in the end, detecting AI text reliably is impossible: arxiv.org/abs/2303.11156 Obvious AI writing is obvious only because it is low-effort & prompting multiple times removes detectability. All detectors have high false positive rates & none work well with GPT-4.

In a new paper showing that AI comes up with more effective prompts for other AIs than humans do, there is this gem that shows how weird AIs are... The single most effective prompt was to start by telling the AI "Take a deep breath and work step-by-step!" arxiv.org/pdf/2309.03409…

I am absolutely outraged to share that one my bosses, Hooded Horse President Snow Rui, got roofied while at GDC. This is her story in full which I'm helping to amplify: linkedin.com/pulse/i-roofie…

RIP to a fount of wisdom, an amazing person “We thought the probability was 0.2,” Treisman told the Philadelphia Inquirer after Dr. Kahneman’s [Nobel] award was announced. washingtonpost.com/obituaries/202…