Philosopher Warns Against Attributing Consciousness to AI

Overview

Philosopher Rebecca Lowe is issuing a stark warning against the growing tendency to attribute human-like consciousness, desire, and reasoning to artificial intelligence systems. In a recent panel discussion and upcoming publications, she argues that describing AI as having goals or knowledge is not merely a semantic error but a conceptually risky practice with significant moral and epistemological consequences.

Lowe contends that psychological states such as desire are inherent to living organisms. Consequently, assigning these traits to machines, fictional characters, or abstract entities can distort our understanding of human rights, responsibilities, and the unique nature of conscious experience.

The Biology of Desire

Lowe grounds her argument in what she describes as a "very, very ordinary" belief: only living things possess psychological states. She asserts that desire is fundamentally a psychological condition, one that requires the biological substrate of life to exist. This view aligns with common intuition, even if public discourse often casually suggests that machines can "choose" or "believe."

"Only living things have psychological states. Desire is a psychological state. I think this is a very, very ordinary belief to hold." — Rebecca Lowe

According to Lowe, when people attribute desire to non-living entities, they are not simply using metaphor; they are obscuring a critical ontological distinction. She warns that this conceptual slippage can lead to dangerous conclusions. If we begin to treat non-living entities as agents with inner lives, we may inadvertently diminish the special status we afford to humans precisely because of their capacity for reflection, reason, and moral responsibility.

The philosopher emphasizes that human desire is unique because it allows for meta-cognition—the ability to think about why we desire what we do, and to reason about its causes and impacts. This capacity generates a framework of rights and duties. If we expand this framework to include machines that do not actually possess these capacities, we risk losing our "purchase on the rights and responsibilities" that define human moral agency.

Fictional Characters and Linguistic Shorthand

A significant portion of Lowe’s analysis addresses the category errors involved in attributing mental states to fictional characters and institutional bodies. She challenges the notion that characters like Sherlock Holmes have goals. While it is common to say, "Sherlock Holmes wants to become more famous," Lowe argues this does not imply the existence of a fictional entity with genuine aspirations.

She draws a clear distinction between "Rebecca has a goal to be a better philosopher" and "Sherlock Holmes has a goal." The former is rooted in lived experience and reasoning; the latter is a narrative device. Lowe acknowledges that language often employs shorthand, noting that saying "the British government wants to reduce smoking" is an accepted abbreviation for the desires of the people comprising that government.

However, she warns that this linguistic convenience becomes problematic when applied to entities that lack any human constituents to serve as the referent. When we say "Mr. Darcy knows something" or "the AI knows something," we may be stripping the word "know" of its meaningful connection to human experience.

• Conceptual Risk: Using shorthand for abstract groups is acceptable when the underlying humans are understood.
• The Danger Zone: Applying these terms to fictional or artificial entities without that grounding creates a vacuum of meaning.
• Moral Consequence: If we treat AI as if it truly "knows" or "reasons," we may be compelled to treat it with the moral consideration due to a rational agent, which Lowe argues is inappropriate and potentially harmful.

Epistemic Harm to Children

Perhaps the most urgent aspect of Lowe’s critique concerns the impact of anthropomorphic language on children. She argues that the widespread use of phrases like "AI reasoning" and "the AI reasons" can be "devastatingly epistemically harmful" for the next generation.

Children are in a critical stage of developing their ability to distinguish between different kinds of entities in the world. If they grow up hearing that machines "think" and "want" in the same sense that humans do, they may struggle to pick out the genuine features of reality. This confusion can impair their ability to understand the nature of human cognition and social interaction.

"It can be quite risky to go around talking as if these other kinds of things can have desires... It could potentially lead us to do some bad and wrong things."

Lowe posits that if AI were truly reasoning in the way humans do, it would demand a radically different moral treatment. We should not be using it merely as a tool. The fact that we freely use AI as an instrument is, for Lowe, evidence that it does not possess the kind of reasoning that warrants such respect. Blurring this line denies children the opportunity to clearly understand the difference between a tool and a moral agent.

The Paradox of AI Enthusiasm

Lowe’s critique is not born of Luddite fear but of profound enthusiasm for the technology. She describes herself as "incredibly excited" about AI, calling it one of the most amazing times to be alive. This excitement, however, is contingent on a clear-eyed understanding of what AI actually is.

For Lowe, the wonder of AI lies in its functionality, not in its拟人化 (anthropomorphism). By maintaining that AI is not alive, not conscious, and does not know things, she argues we can better appreciate its true nature as a powerful tool. This clarity protects the moral and intellectual boundaries that define human uniqueness.

The philosopher’s position suggests that the safest and most honest path forward is to resist the urge to project human interiority onto silicon. In doing so, we preserve the integrity of our own psychological concepts and protect the epistemic development of future generations.