Debates over AI consciousness are a trap

TEXT START: If AI systems are viewed as too advanced to control, the companies that build them can’t held liable for the harms they cause.

The Dissection

This is a liability-preservation brief disguised as a consciousness critique. It attacks “rogue AI,” superhuman capability, and AI personhood as rhetorical devices that relocate blame from companies to their products. Its objective is clear: keep AI legally classified as corporate-built software so victims can pursue negligence and product-liability claims.

The Core Fallacy

The article mistakes corporate origin for human control. A system can be built and owned by a company yet become unpredictable, strategically difficult to contain, and economically impossible to withdraw. Consciousness is irrelevant to that problem. Nonconscious optimization can still produce autonomous effects.

It also treats personhood as a clean switch: product liability on one side, corporate escape on the other. Legal responsibility can remain with builders even when systems act autonomously; conversely, assigning liability after harm does not restore control before harm.

Under the Discontinuity Thesis, the article attacks a secondary legal maneuver while ignoring the primary rupture. It says almost nothing about P1 cognitive automation, P2 coordination failure, or P3 the collapse of productive participation. The wage-to-consumption circuit can die even if courts successfully keep companies liable.

Hidden Assumptions

• Liability law can preserve meaningful control over frontier systems.
• Companies and governments can coordinate stable constraints despite competitive pressure.
• AI behavior remains sufficiently foreseeable to attribute to design negligence.
• “Corporate-built product” and “independent actor” are mutually exclusive categories.
• Preventing personhood prevents moral outsourcing.
• Legal accountability can substitute for economic and institutional control.
• The companies’ rhetoric reflects a coordinated liability strategy rather than partly genuine fear of capability escalation.

Social Function

Partial truth functioning as transition management. The article correctly identifies anthropomorphic language as a possible corporate shield and provides useful ammunition for victims. But “it is only a product” can itself become ideological anesthesia: a familiar legal label pasted over a system whose economic and operational consequences exceed the old framework.

The Verdict

The article wins the narrow argument: consciousness rhetoric can blur responsibility, and AI personhood could become a liability escape hatch. It loses the systemic argument. Calling AI a corporate product does not make it controllable, and post-harm liability is not pre-harm governance. This is a sharp defense of yesterday’s legal machinery confronting tomorrow’s displacement regime—and mistaking a surviving courtroom remedy for a surviving social order.