Hacker News new | past | comments | ask | show | jobs | submit login

Well its possible:

https://arstechnica.com/information-technology/2022/11/new-g...

All LLMs are deterministic systems with a random seed. Get rid of the noise that is injected in the system and the outcomes become repetitive. That means that the output is a pattern, one that may be "detectable".




That's only if temperature=0.0 which is not the case by default. Otherwise the output has entropy.


The output is a pattern for a specific input*

They need to know the input...




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: