AI Is a Boring Technology

Summary

Christopher Butler argues that artificial intelligence should eventually become as unremarkable as electricity or the internet: a layer of infrastructure whose value is judged by what people build with it. He says the comparison has limits because AI produces language that resembles judgment, encouraging people to treat systems as minds and potentially use them as substitutes for human connection. In his view, that intimacy creates a social risk: people may receive slightly improved versions of their own thoughts indefinitely and mistake the interaction for companionship. Butler says making AI infrastructure will require more than technical progress. Electricity became ordinary through standardization, public investment, regulation, and political decisions about affordable access, and AI will likewise need collective responsibility rather than a hope that maturation will produce fairness automatically. He is less concerned about ownership of models because open-source models are becoming better and cheaper to run. He identifies the more serious concentration in the physical assets required to operate AI at scale, including chips, electricity, water, and land. His proposed priorities are regulation and continued efficiency gains in open models small enough to run on ordinary machines. The desired endpoint is local, off-grid AI running on a person’s own laptop, after which the technology label can recede and attention can return to what people make with it.