A look at AI chatbots modeled on real experts, like US psychologist Martin Seligman, that are often created without permission, highlighting a policy gray area
and it only gets weirder from there. https://www.politico.com/... Jason Colavito / @jasoncolavito : I don't suppose it occurred to anyone celebrating the A.I. “replicas” of scholars that draw on their published works and interviews that real people draw influence from knowledge and experience and ideas that they have never committed to paper. https://www.politico.com/... Paul Demko / @pauldemko : “My instinct is strong and immediate on it,” said Tim Wu, a legal scholar who architected the Biden administration's antitrust and competition policies. “I think it's unethical and I think it's something close to body snatching.” https://www.politico.com/... Ryan Brune / @bruneelections : Apparently there is an American psychologist that is a household name in China. He works in the well being / positive psychology space. He has recently been distilled into a chatGPT by Chinese researchers. https://www.politico.com/... [image] LinkedIn: Rafael Brown : This is the problem with AI hype. You get non-technical sites like politico, putting together sensationalist headlines as they describe a thing they literally don't understand at all. …
Context & Ripple Effects
AI replicas extend generative AI from producing generic answers to packaging a recognizable person's perceived expertise into an interface. The core unresolved issue is not simply model accuracy, but whether published work and interviews are enough to support a chatbot presented as that person.
That distinction matters most in high-trust settings: reporting on AI chatbot therapists already shows users seeking always-available conversational guidance, while critiques of ChatGPT’s reliability question the authority users may assign to an expert-branded system.
First-order effects
- Experts such as Martin Seligman can lose practical control over how their name, views, and implied endorsement are presented when replicas are built without permission.
- Developers of expert-modeled chatbots face an immediate policy gray area over consent, attribution, and responsibility for answers issued under a real person's identity.
Second-order effects
- Users may treat responses as expert advice rather than generated output, increasing the importance of clear provenance and limits—especially as AI tools are used for emotionally sensitive guidance.
- Platforms and builders may be pushed to distinguish general knowledge assistants from identity-based replicas, including through consent processes, labeling, or access restrictions.
Third-order effects
- If expert replicas become a common interface, governance is likely to shift from narrow questions of training data toward rules for synthetic identity, endorsement, and accountability for conversational agents.
- The longer-term market could favor systems that can demonstrate authorized identity use and reliable boundaries; the exact legal standard remains unresolved in the coverage.
The trend: This is one data point in the move from general-purpose chatbots toward governed AI companions and expert personas whose usefulness depends on trust and identity rights.