Mark Riedl on Nostr: This paper evaluates LLMs in several search engines. Responses contain unsupported ...
This paper evaluates LLMs in several search engines.
Responses contain unsupported statements and inaccurate citations a majority of the time.
LLM-powered search isn't ready for prime-time.
Seems like a "no duh" sort of result, but the paper needed to be written
https://arxiv.org/abs/2304.09848Published at
2023-04-20 01:53:23Event JSON
{
"id": "216a1688cf0045954a8944cd247978a07d03e18cabfbe246b5d422e7aeb61a1d",
"pubkey": "09a198590597d1d2887bb2a675ed00c30250c483d9bf72b4cb9b8117e6251d1c",
"created_at": 1681955603,
"kind": 1,
"tags": [
[
"mostr",
"https://sigmoid.social/users/Riedl/statuses/110228642419695751"
]
],
"content": "This paper evaluates LLMs in several search engines.\n\nResponses contain unsupported statements and inaccurate citations a majority of the time.\n\nLLM-powered search isn't ready for prime-time.\n\nSeems like a \"no duh\" sort of result, but the paper needed to be written\n\nhttps://arxiv.org/abs/2304.09848",
"sig": "038718b47589152978476b8dfdd70e2197aa1c332aba4b2523523bd508fb9f6ee795233b22d441ab91d7e76670ed5cba99ef70bb20e308659f958fdea32b6196"
}