TrisEck

All the while Karpathy talking about how only a fraction of people understand the capability of (frontier) AI models: https://news.ycombinator.com/item?id=50047422.

As a product manager my approach and execution of my daily work has massively changed in the last half a year. It never stops making me baffled how little people out of software development understand about what is happening right now.

I would love to see some follow up research from OpenAI on some of these hypotheses. While this sounds logical from how humans act, I wonder it the abstraction of the problems still applies to the complex mechanisms and systems built around AI training.