49
ChatGPT Can Be Broken by Entering These Strange Words, And Nobody Is Sure Why (www.vice.com)
submitted 3 years ago by to c/tech@kbin.social
22 comments fedilink hide all child comments
 

Reddit usernames like ‘SolidGoldMagikarp’ are somehow causing the chatbot to give bizarre responses.

you are viewing a single comment's thread
view the rest of the comments
[–] 2 points 3 years 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

  • permalink
  • fedilink
  • source
  • parent