*An LLM can write a poem that fools me into thinking it’s good, but not anyone w... | Hacker News

Hacker News new | past | comments | ask | show | jobs | submit

login

CamperBob2 on July 11, 2023 | parent | context | favorite | on: AI Safety and the Age of Dislightenment

An LLM can write a poem that fools me into thinking it’s good, but not anyone who reads poetry as a hobby.

And you don't grasp the fundamentally-incremental nature of this point?

jononomo on July 11, 2023 [–]

No. After you've asked ChatGPT to write 5-6 poems you'll get a sense for how profoundly uncreative it is as a poet. It's not going to "incrementally" be anything other than a dead-as-a-doornail algorithm.

Aeolos on July 13, 2023 | [–]

That's ChatGPT version 3. Version 4 is already significantly better.

What about version 5? 6? 10?

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact