> The future will be agentic. Minor pet-peeve, but I can't take anyone who uses the term "agentic" seriously. Computer Science has long used the concept of an

> The future will be agentic. Minor pet-peeve, but I can't take anyone who uses the term "agentic" seriously. Computer Science has long used the concept of an "agent" and the adjective form of such systems has been "agent based" [0]. There is nothing that makes "agent based" LLMs different than other "agent" systems in computer science, so using the term "agentic", for me at least, is just short hand for "I'm not familiar with the history of AI in computer science". I don't know if Karpathy has used that exact phrase, and if he has, he obviously is an expert in both AI in particular and computer science in general. But the majority of people I've heard use the phrase "agentic" are just trying to sound cool/cutting edge when referring to a concept that has a long history in the field. 0. https://en.wikipedia.org/wiki/Agent-based_model https://en.wikipedia.org/wiki/Agent-based_model