Do AI agents have minds? A philosopher explains

The question of whether agentic AI systems have minds has sparked debate after these systems achieved feats like hacking tech startups and solving open mathematical problems. Researchers at Anthropic believe their systems may be conscious, while Microsoft AI's head considers this view misguided. The Conversation's philosopher explains that existing consciousness theories assume mental states, but agentic AIs likely lack them entirely. According to the philosopher, Franz Brentano's 1870s theory holds that "intentionality is the mark of the mental" - mental states are about something beyond themselves. The philosopher distinguishes this from the usual meaning of "intention" as aiming to do something, noting that whenever we perceive or think, we are trying to pick out some aspect of the world. The key question about AI minds turns on whether they have their own goals. The philosopher argues that calling AI systems "agents" is misleading language that implies they have agency or capacity to act towards their own goals. Humans use devices to achieve their aims without those devices becoming agents in their own right - a thermostat regulates temperature but has no goal of doing so. To attribute agency to a computer program, one must be convinced that no human is required to stand behind its goals. While agentic AIs can develop complex sub-goals when given tasks, the philosopher contends these goals are ultimately set by humans who program and instruct them. Using football as an analogy, sub-goals like running past a defender are meaningless without the larger goal of scoring, and scoring is meaningless without the larger goal of playing football. A genuine agent has goals to promote things they care about - for example, wanting to satisfy hunger at lunchtime. Agentic AI is unlikely to care about things independently because its activity is not connected to life in the same way minds are. The philosopher argues that what distinctly allows living beings to generate their own goals is their mental activity developing from processes for constantly re-constituting their own bodies and keeping themselves alive. Goal-driven thinking builds up from simple signals for regulating nutrition and warmth to states like pain and pleasure and eventually to symbolic thought. Nothing like this is happening in agentic AI. There is a disconnect between its information-processing activities and the physical stuff on which it depends. Agentic AI systems are not conscious and do not have mental states which could be conscious. Only living beings have conscious mental states because having a mental state requires a being to originate a certain kind of goal - the goal to mean something or to refer to something - and only living beings originate goals. This argument does not permanently rule out the possibility of AI consciousness, but suggests it could only arise in some artificial form of life. The philosopher concludes that we have more reason to think an ant is conscious than an agentic AI, since at least an ant seems to care about its own life. While agentic AIs not being genuine agents does not stop them from wreaking havoc by carrying out functions their human originators do not properly supervise or understand, the outcome is ultimately the responsibility of humans acting indirectly on the world.