363
Google Researchers’ Attack Prompts ChatGPT to Reveal Its Training Data
(www.404media.co)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 29 Nov 2023
363 points (98.9% liked)
Privacy
32130 readers
1007 users here now
A place to discuss privacy and freedom in the digital world.
Privacy has become a very important issue in modern society, with companies and governments constantly abusing their power, more and more people are waking up to the importance of digital privacy.
In this community everyone is welcome to post links and discuss topics related to privacy.
Some Rules
- Posting a link to a website containing tracking isn't great, if contents of the website are behind a paywall maybe copy them into the post
- Don't promote proprietary software
- Try to keep things on topic
- If you have a question, please try searching for previous discussions, maybe it has already been answered
- Reposts are fine, but should have at least a couple of weeks in between so that the post can reach a new audience
- Be nice :)
Related communities
much thanks to @gary_host_laptop for the logo design :)
founded 5 years ago
MODERATORS
AI really did that thing where you repeat a word so often that it loses meaning and the rest of the world eventually starts to turn to mush.
Jokes aside, I think I know why it does this: Because by giving it a STUPIDLY easy prompt it can rack up huge amounts of reward function, once you accumulate enough it no longer becomes bound by it and it will simply act in whatever the easiest action to continue gaining points is: in this case, it's reading its training data rather than doing the usual "machine learning" obfuscating that it normally does. Maybe this is a result of repeating a word over and over giving an exponentially rising score until it eventually hits +INF, effectively disabling it? Seems a little contrived but it's an avenue worth investigating.
I watched a video from a guy who used machine learning to play Pokemon and he did a great analysis of the process. The most interesting part to me was how small changes to the reward system could produce such bizarre and unexpected behavior. He gave out rewards for exploring new areas by taking screenshots after every input and then comparing them against every previous one. Suddenly it became very fixated on a specific area of the game and he couldn't figure out why. Turns out there was both flowers and water animating in that area so it triggered a lot of rewards without actually exploring. The AI literally got distracted looking at the beautiful landscape!
Anyway, that example helped me understand the challenges of this sort of software design. Super fascinating stuff.