PulseAugur
EN
LIVE 09:29:08

User dismisses LLMs as RNGs, citing programming task failures

A user on Mastodon expressed skepticism about the current capabilities of large language models (LLMs), likening them to a random number generator that occasionally produces correct outputs. The user stated that most LLM outputs are "trash" and shared an anecdote about a simple programming task involving multiple if-then checks that even advanced local models fail to solve, citing AI's inability to correctly count brackets as a specific example. AI

IMPACT User expresses a negative opinion on LLM capabilities, highlighting limitations in logical reasoning and counting.

RANK_REASON User opinion piece on social media about LLMs.

Read on Mastodon — sigmoid.social →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

COVERAGE [1]

  1. Mastodon — sigmoid.social TIER_1 English(EN) · [email protected] ·

    Anyone who spent some time with and investigating # LLM based # AI knows that it's nothing other than an RNG that happens to be right sometimes. But most output

    Anyone who spent some time with and investigating # LLM based # AI knows that it's nothing other than an RNG that happens to be right sometimes. But most output is trash anyways. I have a simple programming task that even the best local models can't solve yet. It's about multiple…