2mo ago

I am disappointed in the AI discourse

the top post on my Bluesky feed was something along these lines: "ChatGPT is not a search engine. It does not scan the web for information. You cannot use it as a search engine. LLMs only generate statistically likely sentences." The thing is… ChatGPT was over there, in the other tab, searching the web. And the answer I got was pretty good.

Technology @lemmy.world

anus @lemmy.world

2mo ago

steveklabnik.com /writing/i-am-disappointed-in-the-ai-discourse/

2 comments

Except most of the time it's wrong information, as research has shown. Sounds good though.
I feel like going with the comment in the LW community that the author is directly contributing to the problem. They pick some claims they like and then draw a very under-complex conclusion. What they leave out is that ChatGPT frequently lies to me. With facts, whether it used dynamic memory in the C++ code it just generated for me... And there is a big difference if it pulled the "facts" out of it's model or used Google or RAG. And then we have things like human skill degradation which need further study. Or people planning their vacation with ChatGPT and subsequently needing someone to rescue them in the mountains. And the issue with generative AI art isn't whether some blog author likes the resulting pictures. That's just not what that discussion is about.
So, I'm also disappointed in the discourse. And I'm also often opposed to the stochastical parrot argument, since most people misrepresent the consequences of it.