3 ms·
Any output of the system that interacts with humans can be used for this, it doesn't necessarily need to be robots and physical manipulation. For example, it co
by CGamesPlay 3y ago
Any output of the system that interacts with humans can be used for this, it doesn't necessarily need to be robots and physical manipulation. For example, it could be "code a product" where the models all get the same spec and compete to provide the best implementation, or "create the funniest meme" where the models are judged by each other and then grounded in the real-world rules of reddit upvotes.
- modeless 3y agoAny grounding that comes from humans is scarce. For training we really want automatic grounding from rules that can be applied trillions of times per second rather than labor intensive human grounding at much lower rates. Of course, grounding from robots is also scarce unless you build a whole lot of robots. Ultimately we need systems that generalize rules from a small amount of grounding data. I guess that describes world models, which large language models currently lack (explicitly, at least).
- skybrian 3y agoGrounding from programming languages (or similar sandboxes) wouldn't be scarce and feedback would be faster. It only goes so far, but considering how far LLM's got already, it seems promising.