| ▲ | Ask ChatGPT or Gemini the same "best X" twice, ~1/3 of the answer changes(github.com) | |
| 2 points by EpsilonJ 11 hours ago | 2 comments | ||
| ▲ | apothegm 10 hours ago | parent | next [-] | |
Yes that’s what it means for it’s output to be stochastic. | ||
| ▲ | nosioptar 9 hours ago | parent | prev [-] | |
I've found that asking a clanker something like, "you sure about that? Sounds like you're wrong" will usually get it to change its answer. How people trust any output from these things is beyond me. | ||