ChatGPT is Thinking Fast and Slow
It’s going to cost us — and the planet
When I studied AI and Cognitive Science at Johns Hopkins, Kahneman and Tversky were already gods.
The duo became famous (and went on to win a Nobel prize) for proving something that everyone knew intuitively: people aren’t rational, and they consistently take mental shortcuts and succumb to biases when making important decisions.
Kahneman and Tversky’s early work is legendary in the Cog Sci community. But it was Kahmenan’s bestselling book, Thinking, Fast and Slow, that catapulted the pair’s theory to Oliver Sacks-levels of household fame (Tversky tragically passed away before he could join in the fun of publishing a pop-sci blockbuster.)
Now a decade later, Kahneman’s book and the duo’s theory is influencing the direction of the generative AI revolution, and OpenAI’s latest AI breakthrough.
ChatGPT was already great at thinking fast. With the release of ChatGPT o1 (codenamed Strawberry), the model is now learning to think slow.
Systems of the Brain
Kahneman and Tversky’s theory is popular with laypeople in part because — unlike much writing on neuroscience — it’s easy to understand and has practical lessons for daily life.