Post #4035491

I have been in a few of the many, many conversations this week at #IETF regarding if and how to built new Internet protocols for AI agents, and I have so many questions. So many, but maybe also just one. Are we ultimately moving towards building a separate internet (lowercase on purpose) just for AI? Is that what all these discussions lead to? FWIW, I think that’s probably a terrible idea. I don’t think I’ve heard anything that suggests the problems we’re attempting to solve are specific to agentic AI. AI, as always, makes things go faster (and so scale is a legit problem) but the discovery problems, the identification problems, the authentication and authorization problems… they aren’t precisely new. There are more of them, and there are questions as to whether agents have any right to privacy, but do we need entirely new protocols to account for that? hmmmmmm.