Post #989373
2026-03-26 22:25 UTC
It is very rare for me to search something on the web and get to something as profound as this.
I wanted to know why LLM are based on a random number generator. That’s how I found this article and boy do I understand now.
https://towardsdatascience.com/llms-are-randomized-algorithms/
#ai #llm #openai #anthropic #programming #chatgpt #google #microsoft
Replies (1)
-
@fd93@fosstodon.org 2026-03-26 22:42
@hmiron For me one of the most interesting take aways here is that companies are targeting public perception of LLMs by the adversarial method. I used to work with expert systems chatbots, which are more deterministic. Thus, when they fail they just fail. It's very uh "interesting" that introducing randomness has taken us from Cleverbot, which was a joke, to LLMs, which people consider "intelligent", within 10 years.