The GPT-X models we have seen are not agents. On the other hand they can be used as part of an agent in a system like AutoGPT.
If you treat a language model itself as a dangerous agent you likely mistake a lot of the complexity that goes into the agents that are actually dangerous.
The GPT-X models we have seen are not agents. On the other hand they can be used as part of an agent in a system like AutoGPT.
If you treat a language model itself as a dangerous agent you likely mistake a lot of the complexity that goes into the agents that are actually dangerous.