3 ms·
But can these really be trusted? There was just a HN post which proofed that you can train a model to behave completely different on a certain day. How do we no
by _ink_ 1mo ago
But can these really be trusted? There was just a HN post which proofed that you can train a model to behave completely different on a certain day. How do we now, that these models do not find a way to call home when they see interesting informations (probably irrelevant on a personal level, but corps, government and military might care).
- wonnage 1mo agoChatGPT already notifies the authorities if it thinks you’re doing something illegal. Fable downgrades itself if it thinks you’re doing something even vaguely suspicious.
- chasd00 1mo ago> ChatGPT already notifies the authorities if it thinks you’re doing something illegal. aside from the obvious IP theft problem, it's probably most dangerous for Chinese users outside China to use the Chinese models.
- Yiin 1mo agoAbove average levels of paranoia here, but one way you can prevent that is by not connecting the machine in question to the internet.
- giwook 1mo agoI think it's an average level of paranoia given the level of corporate espionage that happens on a daily basis, and it's not just China that's doing it.