intelligent agent
/ in-TEL-uh-junt AY-junt /
An intelligent agent is the standard way modern AI frames any system that acts on its own: something that perceives its environment through sensors and acts on it through effectors to pursue some goal. A thermostat is a tiny agent (it senses temperature and switches a heater); a self-driving car is a big one (it senses traffic with cameras and acts with steering and brakes); a chess program, a trading bot, and a Mars rover are all agents. The lens is simple but powerful: percepts come in, actions go out, in a continual loop.
Spelling out the pieces helps. The agent is the decision-maker. The environment is everything outside it that it senses and changes. Sensors gather percepts (camera images, temperature readings, the current board); actuators carry out actions (moving a wheel, sending a message, placing a piece). The agent's behavior is governed by an agent program — a function from what it has perceived to what it does next. This framing deliberately says nothing about whether the agent 'thinks'; it focuses on the observable cycle of sensing and acting.
Why this matters: it gives AI a single, unifying vocabulary. Wildly different systems — a vacuum robot, a search engine, a game AI — can all be analyzed the same way by asking what the agent perceives, what actions it can take, what environment it lives in, and what it is trying to achieve. It also raises the natural follow-up question that the next idea answers: among all the actions an agent could take, which one is the right one to take? That is where the notion of a rational agent comes in.
A robot vacuum is a textbook agent. Sensors: a bump sensor, a dirt detector, a battery gauge. Environment: your living-room floor. Actuators: wheels and a suction motor. Agent program: 'if dirt detected, clean here; if bumped, turn; if battery low, return to dock.' Describe almost any AI system by filling in those same four blanks.
Sensors in, actuators out: the perceive–act loop that defines any agent.
'Agent' is a framing, not a claim of intelligence — a thermostat counts. The recent buzzword 'AI agent' usually means a language model wired up to take actions (call tools, browse, run code); it is the same perceive–act idea, with all the same fallibility of the model inside.