Corey_

6 ms·

1. 1.▲
   by Corey_ 1y ago

   What’s more worrying isn’t that they tested—it’s that the existence of the test feels like a smoking gun. That says more about our broken trust in institutions than anything else.
2. 2.▲
   by Corey_ 1y ago

   Humans may not update like Bayesians, but we read context, shift priorities, and act under pressure. Judgment isn't just math — it's lived experience, intuition, and meaning in motion. That’s still hard to replicate.
3. 3.▲
   by Corey_ 1y ago

   I completely agree that TypeScript is ideal for LLMs. The type system and the extensive training data make it the best choice. But as someone who's been working with TypeScript for a while, I still see LLMs struggling with complex gene