True agentic AI is nowhere even close to a reality.
By “agentic AI” I mean cognitive based output and decision-making.
That type of output is antithetical to the way current AI works.
Agentic AI would require a machine learning model that can ignore its node weights, while still producing coherent responses.
Personally, I think that capability is impossible until organic machines are developed, and even then it might still be impossible.
Judgement and creativity require being able to rationally say No to your internal “programming”. Humans can do this. Animals maybe can do this (but I don’t think they can). Machines never do.
That’s not to say that AI “agents” can’t be useful tools. They can and will. They just won’t be truly agentic.
RE: It only takes one