Part 2/10:
Today, AI agents are envisioned as capable entities that can initiate actions, perform tasks autonomously, and contribute meaningfully to problem-solving. Traditional AI applications, like ChatGPT or Gemini, operate in a reactive manner—they respond to inputs but lack the ability to act independently. DJ argues that for AI to reach a level of enhanced productivity, it requires a sense of agency, or the ability to act on its own volition.
However, while the notion of agentic AI is exciting, the definition is fraught with inconsistencies. Some products may claim to be AI agents based on various criteria, but to truly embody the concept, they should meet four key characteristics laid out in the 1995 paper "Intelligent Agents: Theory and Practice." These characteristics are: