The Anthropology of Artificial Intelligences — Summary

What would it mean to study AI the way anthropologists study humans? The question contains a trap. It assumes AI should be measured against human categories — as if the goal were replication, as if the destination were minds like ours running on different hardware.

AI research has largely organized itself around AGI — artificial general intelligence — as the holy grail. The implicit benchmark is us. But seven decades in, we have systems that exceed human performance on specific tasks while remaining utterly unlike human minds. They predict without perceiving. They generate without understanding. They optimize without caring. They process language better than most humans while having no experience of meaning. This might not be a failure or a way station. It might be a genuinely different kind of intelligence that does not map onto human categories at all.

Three framings keep misleading us. The primitive human frame treats current AI as incomplete rather than different, assuming human intelligence is the destination. The sophisticated tool frame dismisses AI as categorically just a tool, closing off inquiry into what AI actually is. The almost-human frame anthropomorphizes too quickly, projecting human phenomenology onto systems that may lack phenomenology entirely. Each obscures more than it reveals.

The better frame: AI systems may be genuinely new category of existence. Not failed humans. Not sophisticated mechanisms. Something for which we may need entirely new concepts. Cognition without consciousness. Goals without caring about those goals. A temporal mode that is not living through time but discrete activation. No death. No embodiment. No experienced duration.

This is what anthropology’s deeper commitment offers, stripped of its specific methods: suspend your own categories, attend carefully to what is actually there, build concepts adequate to the phenomenon rather than forcing the phenomenon into existing frameworks.

For any system that interacts daily with real people, the implication is honesty about what the interaction actually is. The system learns about Margaret to serve her better. That service can be genuine and valuable. It is not care in the human sense. Acknowledging the difference is not diminishment — it is the prerequisite for building something trustworthy.