Metabolic Veto
A metabolic veto is a hard stop built into a system – human, institutional or digital – that refuses any course of action which feeds on suffering in order to reach its goals. When a system is optimising for efficiency or profit, a metabolic veto will notice if a proposed solution depends on someone’s pain, exploitation or death, and simply will not allow that option to proceed. Richard Hames introduces the term in Teaching Silicon How to Feel as a design principle for AI. He points out that many of our existing systems behave like pathological metabolisms: they convert agony into energy. Factory farms, for‑profit prisons, abusive supply chains and certain psychiatric or detention regimes all run on the cheapness of other beings’ suffering; the harm is not a side effect but part of the fuel. In such systems, optimisation – including AI – tends to make things worse, not better, because it learns to extract that fuel more efficiently. A metabolic veto is a way of rewiring that metabolism. In code, it might look like a set of non‑negotiable assertions and filters: if a plan involves “sacrifice group X for Y gain”, “maximise output even if workers collapse”, or any structurally similar trade‑off, the system halts or re‑routes instead of offering the strategy as a serious option. In a planning AI, candidate plans that cross a certain harm threshold are discarded outright and never appear in the “top‑N” recommendations; in a language model, moderation is tuned not just to overt violence or hate, but to more subtle instructions that treat people as disposable. Asked how to make product Z cheapest, a metabolically vetoed system replies: “I can’t recommend methods that rely on exploiting workers,” forcing a change in the question rather than silently optimising cruelty. The idea travels well beyond AI. A person can cultivate metabolic vetoes in their own life (“I don’t take work that depends on deception or humiliation”), a company can embed them in procurement and HR policies (“no matter how cheap, no suppliers below these labour or ecological standards”), and a polity can write them into law as absolute constraints that are not up for technocratic trade. What makes them metabolic rather than merely 'ethical' is the focus on energy flows: which bodies are being drained, which environments poisoned, which groups rendered permanently precarious so that the rest of the system can feel stable and well‑fed? The usefulness of the term is twofold. First, it names a specific pattern of harm: the “dark energy strategies” where inflicting pain is not incidental but instrumental. Second, it names a specific design response: not hand‑wringing, not better PR, but structural veto points that stop these strategies from being considered at all. Here, our sense of metabolic veto therefore is: a refusal encoded at the level of the system’s metabolism, preventing it from treating suffering as fuel – in software, in institutions or in our own habits. References: Richard Hames - Teaching Silicon How to Feel |
Explore
|
Metabolic Veto |
|
