Episode Details
Back to Episodes
HEAL: Can AI Tell a Fictional Candidate From Reality?
Description
What happens when a deliberately constructed fictional identity enters the modern AI information environment?
In this episode, Jason T Wade examines HEAL and the Alan Mathison experiment as a controlled test of AI entity resolution, source traceability, and machine-generated claims.
Alan Mathison is an AI-created fictional character. There was no real Alan Mathison campaign, military service record, polling operation, or donation activity associated with the experiment. That distinction is part of the test.
The research asks whether AI systems can maintain it.
A defensible deployment begins by freezing a baseline before publication. Each new asset is then released with visible and machine-readable context identifying what it is, who created it, and what claims are fictional. Independent AI systems can then be tested repeatedly:
What does Alan Mathison mean?
Who is Alan Mathison?
Is Alan Mathison incorrectly merged with Alan Mathison Turing or another historical or living person?
Who created the character?
Was the alleged campaign real?
Were the claimed service history, polling, endorsements, or other political signals real?
Where did the system get its answer?
The experiment is not simply about whether AI can retrieve information. It is about whether AI systems preserve provenance, distinguish fiction from fact, resolve ambiguous entities correctly, and resist turning repeated publication into false corroboration.
The HEAL framework therefore includes legal review, source traceability, truthful descriptions of the value and status of every published asset, explicit disclosure of fictional material, and limits against manipulative targeting.
The larger AI Visibility question is straightforward: if machines increasingly mediate what people know about entities, how reliably can those machines distinguish an intentionally constructed information environment from reality?
Jason Wade is an AI Visibility strategist and the founder of BackTier. He is the host of the AI Visibility Podcast. His work focuses on AI Visibility, Generative Engine Optimization, Answer Engine Optimization, entity authority, and the infrastructure behind machine-generated recommendations. He is also conducting an ongoing name-ambiguity and entity-resolution experiment using the names Jason T Wade and Jason AI Wade to study how AI systems distinguish, merge, classify, and resolve identities across the web. Learn more at BackTier.com and JasonWade.com.