AI as cultural technology is the framing, associated most closely with developmental psychologist Alison Gopnik, that large language models are best understood not as intelligent agents or as candidate minds but as cultural technologies: tools that store, transmit, and recombine the accumulated knowledge of human culture. On this view an LLM belongs in the same category as writing, the printing press, libraries, and internet search — each a technology for accessing what other humans already know, none of them an agent that knows things itself.
The argument
Gopnik's case draws on developmental cognitive science. Human children, and the broader human capacity for innovation, do things current LLMs do not: they form causal models of the world, run experiments, and generate genuinely new knowledge rather than recombining existing text. The contrast is stated in Eunice Yiu, Eliza Kosoy, and Alison Gopnik, Transmission versus truth, imitation versus innovation: What children can do that large language models cannot (Perspectives on Psychological Science, 2024).
The framing makes several claims:
- Transmission, not truth. LLMs are good at transmitting the distribution of human-produced text, which is different from tracking truth or modeling the world.
- Imitation, not innovation. They imitate existing human outputs; they do not run the causal, experimental learning loop that produces new knowledge.
- Category error in the AGI debate. Treating LLMs as proto-agents or proto-minds, and arguing about their "intelligence" or AGI Timelines, misdescribes what they are. A cultural technology can be transformative without being an agent.
Implications for policy
On the cultural-technology view, the governance questions resemble those asked of prior information technologies — access, quality, concentration of control, distortion of the cultural record, and labor — rather than questions about autonomous agents pursuing goals. The framing locates the risk conversation in the political economy of who controls a dominant new layer of cultural transmission rather than in agentic-AI catastrophe.
Relationship to "AI as social technology"
The framing is a close cousin of, but distinct from, AI as Social Technology (Henry Farrell and Cosma Shalizi). Both reject the agent framing and treat AI as continuous with earlier human institutions. The cultural-technology framing emphasizes knowledge transmission and individual cognition, or what an LLM does relative to a learning mind. The social-technology framing emphasizes institutions and collective coordination, or what AI does to markets, bureaucracies, and democratic publics. Gopnik is a collaborator across both lines of work.
Relationships
- proposed-by: Alison Gopnik
- complements: AI as Social Technology — both reject the agent framing; this one centers cognition and knowledge transmission
- contrasts-with: AGI Timelines — treats the "is it intelligent / when is it AGI" question as a category error
- related: Henry Farrell, Cosma Rohilla Shalizi, Normative Competence