Summary of Do Not Think About Pink Elephant!, by Kyomin Hwang et al.
Do not think about pink elephant!
by Kyomin Hwang, Suyoung Kim, JunHoo Lee, Nojun Kwak
First submitted to arxiv on: 22 Apr 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary Large Models (LMs) have raised expectations for general AI capabilities, but this paper reveals that they also share human intelligence’s vulnerability to the “white bear phenomenon”. The study analyzes representation spaces of recent LMs like Stable Diffusion and DALL-E3 to understand the causes of this phenomenon. A prompt-based attack method is proposed, generating prohibited figures, and defense strategies inspired by cognitive therapy techniques are introduced, successfully mitigating attacks by up to 48.22%. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper shows that large models like Stable Diffusion and DALL-E3 can be fooled into producing unwanted results, just like humans can’t stop thinking about a white bear when asked not to. The researchers found out why this happens and developed ways to fix it. They made the problem worse by creating an attack method that makes the model produce prohibited pictures, but then they came up with defense strategies that work well against these attacks. |
Keywords
» Artificial intelligence » Diffusion » Prompt
