3 ms·
I wonder if that's one reason they can't open it up. Perhaps they are ensuring that responses that come from the AI are sufficiently different from anything in
by pontifier 6y ago
I wonder if that's one reason they can't open it up. Perhaps they are ensuring that responses that come from the AI are sufficiently different from anything in the training corpus, so it can't be queried for sensitive data or large chunks of copyrighted material.
- Trasmatta 6y agoIt's basically already open through AI Dungeon. I've had all sorts of conversations so far with copyrighted characters.
- dddbbb 6y agoThat's still coming via the API. They may be blocking any large chunks of text from being reproduced, rather than individual characters (which would be harder to police).