4 Comments
User's avatar
Rainbow Roxy's avatar

Fascinating. I really resonated with Liu’s point that true translation is fundamentally an act of interpretation, not just a one-to-one correspondence. As someone who teaches CS, it makes me think about the inherent limits of current AI models. Can they ever grasp that depth? I'm open to being surprised.

Expand full comment
Evan McLaren's avatar

I didn't expect to read such an AI-positive post today. Food for thought.

Expand full comment
Nils Gilman's avatar

It’s not really AI positive so much as it is AI realist: people are using these platform this way, and so they should get privilege to do so safely from the government. Whether it’s a good idea to do so is another matter: a lot of people are skeptical of therapists and priests, too.

Expand full comment
Giampiero Campa's avatar

To me, the fact that LLMs work so well means that language, as tool to consistently represent reality, is actually better than we originally thought.

Expand full comment