Teachers Teach Students To Spot Anthropomorphism in AI

A TeachThought piece contributed by Dr. Athena Stanley proposes classroom strategies to help K-12 students recognise anthropomorphism in AI. The article recommends starting with everyday examples - naming cars, talking to pets, describing the weather, or saying a computer "hates" them - to prompt questions such as "Why do people do this?" and "When could it become misleading?" The article highlights that conversational interfaces, AI companions, virtual characters, and large language models can make interactions feel personal and blur the line between simulated responses and authentic human communication. The author frames this work as part of foundational AI literacy.

The classroom concern is timely. A June 24, 2026 Brookings Institution piece by Rebecca Winthrop documents how teachers are raising alarms at institutional scale: the Alberta Teachers' Association passed a directive for 50,000 educators barring AI tools that "simulate friendship, counselling, or intimate relationships" from K-12 settings. Winthrop reports the directive covers three categories of anthropomorphic AI: simulations of relationships, simulations of historical figures, and simulations of living people. The Brookings piece also notes that other jurisdictions are moving similarly: China published rules in April 2026 prohibiting services that simulate intimate relationships with minors, and the UK's Department for Education guidelines state developers should not "anthropomorphise products or create products that imply emotions, consciousness or personhood." A US bipartisan bill (GUARD Act) proposes banning AI companions for children and requiring bots to disclose they are not human.

Research cited in Educators Technology (Dr. Med Kharbach, May 2026) explains why the tendency runs deep: LLMs output statistically likely word sequences, not facts about the world. Using words like "believes" and "thinks" to describe LLM outputs, as philosopher Murray Shanahan argues, is a category mistake that actively encourages anthropomorphism. For AI practitioners, the same language issue appears in UX copy, documentation, and prompt design: setting accurate expectations about model capabilities requires deliberate word choice.

Whether K-12 curriculum frameworks adopt explicit anthropomorphism modules, and how edtech vendors respond in product copy and affordances to reduce misleading personification cues.