Recent searches
Search options
“If, as they say, AI tools are going to get better quickly, then let them do so and trust that smart people will pick them up and use them.” From @anildash
https://www.anildash.com/2025/04/19/ai-first-is-the-new-return-to-office/
“If you think your workers and colleagues are too stupid to recognize good tools that will help them do their jobs better, then... you are a bad leader and should step down. Because you've created a broken culture.”
From me: The more people lecture me about how I have to take LLMs seriously because if I’m missing out if I don’t, the less convinced I am that they’re worth taking seriously.
This follow-up post, also by @anildash, imagining a better world, is also excellent
https://www.anildash.com/2025/05/01/what-would-good-ai-look-like/
Even though I think I would prefer to frame this as what “good-enough” LLMs would look like. And acknowledging that clearly these LLMs are good enough for a lot of people and for a lot of their uses. However, they are not good enough for me, and long history suggests that I have reasonably common taste in these things.
Like, a good-enough LLM still needs to have a lying rate that is materially near-zero.
@kevinriggle proposing a new standard evaluation criteria for LLM systems, the Rate Of Flat-out Lying, or ROFL for short
@flaki, criterion, surely…