The Alien Intelligence Thesis

In the AI Story

The key feature that makes AI alien, in Daston's precise sense, is that it is not a social agent. Human knowledge production is embedded in a web of relationships, institutional accountabilities, biographical specificities, and moral commitments that together constitute the infrastructure of trust. A human expert's authority derives not only from the content of her claims but from her location in networks of institutional affiliations, professional reputations, social obligations, and personal stakes that provide the context within which her claims can be evaluated. The evaluative framework developed for human knowledge production assumes this infrastructure — assumes that the knowledge producer is an entity embedded in relationships of trust and accountability, capable of being held responsible for the reliability of its claims, subject to the norms and sanctions of the communities in which it operates.

AI is not embedded in this web. It cannot be held responsible in any meaningful sense. It is not subject to the moral economy of any knowledge-producing community. It operates outside the relationships of trust and accountability that the evaluative framework of human knowledge production presupposes. It is alien not because it is hostile or incomprehensible but because it operates according to principles categorically different from those that govern human knowledge production — and the evaluative frameworks developed for human knowledge do not map onto it without distortion.

The implication is that the calibration challenge is harder than the historical pattern alone suggests. The pattern predicts that evaluative institutions will eventually be built. But the pattern is based on transitions between technologies that amplified human capacities — technologies whose evaluative frameworks could be developed as extensions of existing frameworks. If AI represents an alien intelligence rather than a more powerful version of human intelligence, the evaluative frameworks must be built not by extension but from the ground up, addressing the specific characteristics of a knowledge-producing process that has no precedent in the history of human epistemology.

The alien-intelligence framing also illuminates a specific ethical question that the AI transition raises: whether and how to incorporate entities that cannot be held accountable into the moral economy of human knowledge production. The options are limited. One can exclude such entities, refusing to treat their outputs as knowledge at all — an approach that sacrifices genuine benefits. One can incorporate them as tools whose outputs are evaluated by the humans who deploy them — an approach that places the full evaluative burden on users who often lack the competencies the burden requires. One can build new institutional structures that hold the humans responsible for AI deployment in ways analogous to how the moral economy of science holds scientific authors responsible for their claims — an approach that requires institutional construction unlike anything the present moment has begun.

Origin

The alien-intelligence framing emerged in Daston's public commentary around 2022, building on her earlier work on the historical specificity of objectivity, the moral economy of science, and the genealogy of data. The formulation has been influential in subsequent philosophical discussions of AI, particularly in work that questions the adequacy of extending human-derived evaluative frameworks to AI-generated outputs.

The framing has affinities with earlier arguments by philosophers including Hubert Dreyfus and John Searle about the categorical differences between human intelligence and computational systems, though Daston's approach differs in its historical rather than phenomenological or analytical orientation. What her framework adds is the specific attention to the institutional and moral infrastructure of human knowledge production — infrastructure that AI, as an alien intelligence, does not share and cannot be assimilated to.

Key Ideas

AI is not an amplification of human capacities. Previous knowledge technologies extended human perception or cognition; AI operates according to principles categorically different from human knowledge production.

Social embeddedness is definitional for human knowledge. Human expertise is embedded in relationships of trust, accountability, and moral economy; AI is not embedded in any comparable infrastructure.

Evaluative frameworks must be built from the ground up. Extending human-derived frameworks to AI produces systematic distortions because the frameworks assume infrastructure AI does not possess.

The accountability question has no easy answer. AI cannot be held responsible in any meaningful sense; the responsibility must fall on the humans and institutions that deploy it.

The historical pattern may apply incompletely. Previous transitions developed evaluative frameworks by extension; the AI transition may require construction of frameworks without close precedent.

Debates & Critiques

The alien-intelligence thesis has generated substantial debate. Critics argue that it overstates the categorical difference between AI and human intelligence, treating current AI systems as more different from human cognition than they actually are. Defenders respond that the difference in question is not about cognitive mechanism but about institutional embeddedness — that even if AI and human intelligence shared certain computational properties, the social infrastructure within which human intelligence operates would remain absent from AI operation. A related debate concerns whether the thesis licenses insufficient caution about AI risk (by framing AI as merely different rather than potentially dangerous) or excessive caution (by exaggerating the strangeness of systems that share substantial functional properties with human cognition). The productive position, consistent with Daston's own methodological caution, is that the thesis identifies a specific structural feature that requires specific institutional response, without predetermining the shape that response should take.