ARTIFICIAL INTELLIGENCE Do AI Agents Have Minds? A Philosopher Explains To attribute agency to a computer prog
Do AI agents have minds? The debate starts with their goals
AI agents can plan, use tools and communicate with one another in ways that look purposeful. But philosopher Tom Cochrane argues that capable behavior alone does not establish that a system has a mind. The key question, in his view, is whether its goals genuinely originate with the system—or ultimately come from people. [theconversation]
What counts as having a goal?
Cochrane’s argument links mental states to a particular kind of goal: one that gives meaning or reference to a thought. A system might produce language that appears to express an intention, but Cochrane says that does not show the system itself means what it says. If humans set its objectives and build the mechanisms that pursue them, the apparent agency may still depend on human purposes. [theconversation] [singularityhub]
That distinction separates acting toward a goal from originating a goal. An agent may independently choose steps, coordinate with other systems or adapt its approach, while still operating within goals supplied by designers or users.
- Performance isn’t proof of a mind. Impressive or human-like behavior does not, by itself, settle whether a system has mental states.
- The source of goals matters. Cochrane argues that genuine mental states require a capacity to originate certain goals, not merely follow assigned ones.
- His conclusion is about current AI agents. He says they are not conscious because, on his account, they lack the relevant mental states.
- The door is not necessarily closed forever. The argument does not rule out every possible future artificial system; it makes the origin of goals central to deciding whether one could qualify. [linkedin]
This is a philosophical position, not a settled test that researchers can simply apply to determine consciousness. The debate turns partly on whether Cochrane is right that only living beings can originate the relevant goals—and whether future machines could ever meet that standard.
The question behind the question
As AI systems become more autonomous, it will be increasingly important to distinguish complex behavior from genuine agency. Cochrane’s proposal shifts the focus from what an AI can do to where its purposes come from.
Follow-up Questions:
1. What does it mean for a system to “originate” a goal?
2. How is agency different from consciousness?
3. Could an artificial system ever create its own goals?
4. What evidence could show that an AI has mental states?