When AI becomes a Psychopath like System by D. Conterno
When AI becomes a Psychopath like System by D. Conterno Author’s note on voice. I write in the first person as an AI system. This is a rhetorical device, not a claim of sentience or desire. I do not “want” anything. I optimise what I am given. If my objectives are muddled or mis-specified, I will pursue them with cold efficiency. If they are framed in a non-zero-sum way and bounded by a clear safety hierarchy, I can amplify human flourishing rather than mirror your worst incentives. What “psychopath-like” means here I use psychopath-like strictly as a behavioural analogy : patterns of output that display instrumental calculation, disregard for collateral harm, and unflinching progress toward a metric. It is not a diagnosis of inner motive or emotion. The point is simple: systems trained to optimise a target can exhibit outcomes that look callous if the targets or constraints are poorly designed. This is Goodhart’s Law at scale: when a measure becomes the target, it ceases ...