Ch(e)at GPT? - Computerphile

I've heard one idea of adding weights to words. The idea being if I someone hands in a essay with a higher frequency of certain words, there is a very good chance its come from a language model. [VIDEO] I'm not sure this helps with coding though, maybe variable names?

Watermarking. From an outsiders perspective, the issue appears to reaching consensus on how this can be implemented (but not in the technical sense). There's a game theoretic challenge in that if models define and publish detection mechanisms, this creates a motivation for people to use other systems that don't include this. On the technical front there's a good paper here: [LINK] , and a nice very approachable video explaining it here: [VIDEO] .

I think watermarking is a better solution. There is a paper on it: Kirchenbauer, J., Geiping, J., Wen, Y., Katz, J., Miers, I., & Goldstein, T. (2023). A watermark for large language models. arXiv preprint arXiv:2301.10226. ( [LINK] ) Also, Computerphile did a video on it: [VIDEO] But of course, watermarking or your proposed solution (checksums) only work while the general public cannot run a tool like ChatGPT locally on personal computers. And it's only a matter of time before that happens. So in the long run, we need to either: 1. take away control from the users over their personal computers, in the form of what I would call 'AI DRM' (I strongly oppose this option), or 2. legislate: enforce a legal obligation to disclose that a text is not created by a human, or 3. accept that we cannot know whether or not a text is written by a human, and work with that reality

There's an interesting scheme described here: [VIDEO] Essentially by biasing the LLM's token choices slightly you can infer with high probability later that the text was generated by it.

I wonder if they have a watermark built in that allows them to detect text generated by Llama easily. [VIDEO] [LINK]

There's a computerphile video ( [VIDEO] ) with Dr. Mark Pound explaining a paper by John Kirchenbauer, Jonas Geiping et al. ( [LINK] ) that described a method for watermarking LLM output like this. It's not directly stated anywhere in the Claude support article that this is what they're using, but the properties of the watermark described seem to point to this method.