4 ms·
There’s an enormous opportunity and need for adverse products to quickly detect LLM-produced content. We also need labeling, like the nutritional information o
by aj7 4y ago
There’s an enormous opportunity and need for adverse products to quickly detect LLM-produced content.
We also need labeling, like the nutritional information on food packages.
- scrollaway 4y agoThis idea reminds me of DRM. It’ll always be a game of cat and mouse. And when AI content is too difficult to distinguish from human content you’ll be rife with false positives. It sounds pretty fucking dystopian to me that I would get, for example, banned from commenting somewhere because I didn’t structure my thoughts exotically enough to not be possibly machine-generated. Mark my words: well intentioned as they might be, businesses being created today to detect ChatGPT stuff will in a few years be scummy as hell; you just have to look at the student anti cheat industry. In fact all of these types of businesses end up slimy. DRM, antivirus, anticheat, AML… and now, anti-LLM.