OpenAI recently released the new ChatGPT Agent (previously referred to via the code name “Operator”). It’s pretty wild.
I recently made post about Agent, in which it tries (and fails) to convince an AI that has decided that humanity needs to go.
While that was entertaining and humorous, the release of intelligent Agentic systems is a serious development, and has not been fully thought-out. For example, I was able to have the agent spin up it’s own agent, and when it did so it came up with the task for it’s sub-agent completely on it’s own. In another instance, I got the Agent to make a few Custom GPT’s for me. Check out the video below to see it doing so, and ask yourself “are we really prepared for this?”.
AI Reflections 