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.

Stop Treating AI Models Like People (2023)

17 pointsby laurexover 1 year ago

5 comments

ggmover 1 year ago
This deserves wider distribution. It needs saying constantly. Overattribution is joining my discourse on this. All the language chosen by ai proponents favours belief on a mind. Choice, says, infers, states, hallucinated, thinks.. these are crap words to use describing the systems behaviour of an llm or gpt.
评论 #39609784 未加载
dkjaudyeqooeover 1 year ago
Let&#x27;s start by calling &#x27;hallucinations&#x27; errors.<p>Or we could meet halfway and just call them lies.
htkover 1 year ago
People love to tell others what to do, what to believe and what to think.
评论 #39609913 未加载
johneaover 1 year ago
It&#x27;s such a relief to see someone putting this into print.<p>Now, if we can have the same conversation about pets, we can really make some headway in resolving a major component of first world mental illness.<p>Overattribution and projection...
Zetobalover 1 year ago
Eh... most llms work better if you social engineer them.