Agentic AI

Agentic AI

AI systems, often called "AI agents", designed to pursue a goal with some degree of autonomy: planning steps, using external tools, and taking actions (such as browsing the web, writing and running code, or booking a service) with limited human oversight at each step. Agentic AI raises distinct accountability questions, since the chain of decisions leading to an outcome may not be easily traceable to a single human choice. (See also: Large language model, AI alignment)