TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

What Can LLMs Never Do?

6 pointsby worldvoyageurabout 1 year ago

2 comments

o_nateabout 1 year ago
It's a good article. It seems like the author hedges his bets a little bit on the title question. At the end he allows that someday there could be a combination of LLMs, support scaffolding, and prompt engineering that will be able to tackle even the problems it has difficulty with now. But it was interesting as a tour through some seemingly straightforward problems that even the most advanced current LLMs are bad at, as well as containing interesting speculation on the root cause.
srbhrabout 1 year ago
It is an interesting article, and a good question, "What can AI never do?" There is an ongoing debate on LLMs and Trust, that we should take a look at as well. How much we can trust LLM and the data to make decisions. :)