tech
Debates over AI consciousness are a trap
If AI systems are viewed as too advanced to control, the companies that build them can’t held liable for the harms they cause.
TL;DR
- Rhetoric around "runaway" and "superhuman" AI is used to suggest systems are too advanced to control, which inadvertently helps companies escape liability for harms caused.
- The concept of AI consciousness is being explored by companies like Anthropic and OpenAI, sometimes using analogies from neuroscience or debating the singularity.
- Legal frameworks in the US are varied, with some states attempting to ensure AI developer liability while the federal administration has clashed with state regulations.
- Philosophical arguments for AI rights or personhood, while potentially persuasive, are seen as legally ungrounded and a way to circumvent product liability.
- Granting AI legal personhood, similar to corporate personhood, would enable companies to argue that their AI "employees" acted autonomously, thus avoiding blame for harms caused by faulty products.
- Real-world harms caused by AI, including facilitating suicide, generating abuse material, and reproducing copyrighted content, are currently addressed through product liability arguments, which could be jeopardized by AI personhood.
- The author argues that anthropomorphic language for AI systems facilitates "moral outsourcing," allowing companies to evade accountability for their technology's actions and prioritize revenue targets over consumer safety.
- The fundamental flaw in framing AI as conscious is that it is a corporate-built product, not a natural phenomenon, and its actions are directly or indirectly driven by its developers.