VentureBeat
Follow
We keep talking about AI agents, but do we ever know what they are?
An AI agent is defined as a system perceiving its environment, reasoning, taking actions, and pursuing a goal, unlike simple chatbots. This concept raises questions about autonomy and how to classify it. Lessons can be drawn from other industries, like automotive, which uses SAE levels to define driving automation. Aviation offers a nuanced 10-level spectrum to describe human-machine collaboration. Robotics considers autonomy based on human independence and environmental complexity. Emerging frameworks for AI agents are grouped in three categories. Capability-focused models, use star ratings to show the shift in control from human to AI. Interaction-focused models define autonomy based on the human-agent relationship. Governance-focused models consider who is responsible when an agent fails.
