ChatGPT Can Be Broken by Entering These Strange Words, And Nobody Is Sure Why
Borg286 @ Borg286 @kbin.social Posts 0Comments 21Joined 3 yr. ago
Borg286 @ Borg286 @kbin.social
Posts
0
Comments
21
Joined
3 yr. ago
Interesting that a mere 5 million samples could train ChatGPT so strongly. I see attack vectors.
I predict that like Akinator the beginning will be smart till the internet trolls feed it garbage. From there on put it will be severely flawed