Learn it → Apply it → Save as Skill
You read a story in your own field and catch it getting the facts wrong. You turn the page to a subject you don't know, and read it as if it were reliable. The source did not get more careful. Your trust just reset the moment the subject changed.
The Dr. Wallace’s Favorite Mental Models series is designed to turn useful ways of thinking into reusable tools. Each model gives you a fast visual overview, deeper interactive learning, practical application exercises and an AI-ready Markdown skill you can save for future use.
Instead of merely reading about mental models, you can gradually build your own AI skills library: a collection of portable reasoning skills that can help ChatGPT, Claude, Grok and other AI assistants apply the mental models you value.
Inside your own expertise, you read with a sharp eye and spot the errors. Outside it, on the very next page of the very same source, the sharp eye switches off and you read as a trusting layperson.
Important: none of the source's actual accuracy changed across these four bars. Only your ability to check it did. The gap between the first bar and the last one is the effect itself.
The model is most useful right after you catch a source getting your own subject wrong, and are about to read on to a subject you don't know.
The Gell-Mann Amnesia Effect does not tell you a source is wrong. It tells you that the trust you already withdrew from it should not have reset itself at the page break.
Catch the moment you read something in your own field and spot a real, checkable error, not just a simplification you dislike.
Say it plainly: "this source got my subject wrong," rather than letting the moment pass as a one-off slip.
Assume the error rate you just witnessed travels with the source into subjects you cannot personally check.
Apply the same scrutiny you used in your own field, such as checking primary sources, to the unfamiliar claims that follow.
Open a use case to see what the model changes.
Keep reading the rest of the issue, but lower your default trust in that outlet by roughly the error rate you just measured, and look for a second source before treating any of its other claims as settled.
The same model just produced that confident, well-formatted answer using the same process for the next topic. Carry the same skepticism forward, and ask the AI to cite sources you can actually check for claims outside your expertise.
Resist treating that single confirmed fact as proof the whole document is solid. Spot-check a second, unrelated claim before extending trust to the parts you cannot independently verify.
Getting dates or titles right does not validate subjective character claims from the same conversation. Weight verifiable and unverifiable claims separately rather than letting one accuracy carry the other.
Being right about a sector you know does not mean the analyst's calls on sectors you don't know deserve the same confidence. Check their track record in each sector separately.
Note the error, then read the next chapter with the same attentiveness rather than the usual deference students give printed material.
This is an illustrative example designed to teach the method.
Expand for where the term came from and how it has aged since 2002.
Novelist Michael Crichton introduced the term in a speech titled "Why Speculate?", delivered at the International Leadership Forum in La Jolla, California, on April 26, 2002. He described the experience of reading a newspaper story in his own field, show business, and seeing the journalist get the facts wrong, then turning the page to a subject he knew less about and reading it as reliable.
In Crichton's own words, he named it after physicist Murray Gell-Mann "because I once discussed it with Murray Gell-Mann, and by dropping a famous name I imply greater importance to myself, and to the effect, than it would otherwise have." Gell-Mann did not originate or publish the concept himself; the naming was Crichton's own acknowledged rhetorical move.
Journalist Erwin Knoll is often credited with a related, earlier observation sometimes called Knoll's law of media accuracy: that the accuracy of any news story, as judged by a reader, falls as the reader's own direct knowledge of the subject rises. The Gell-Mann Amnesia Effect adds the specific twist that readers then forget this calibration exists once they move to unfamiliar territory.
The term itself comes from a public speech, not a published psychology study, so treat it as a sharply named observation rather than a formally measured effect with its own citation record. It overlaps with documented research on source-monitoring errors, overconfidence outside one's expertise, and domain-specific calibration, which is why it has held up well as a practical heuristic even without its own dedicated literature.
Treating one caught error as proof an entire source is worthless pushes past calibration into "trust nothing," which is its own failure to reason carefully.
A vague feeling that "something seemed off" does not carry the same corrective power as explicitly stating the specific error you caught.
You can also be wrong about your own subject, or misread a deliberate simplification as an error. Confirm the mistake before generalizing from it.
One caught error should lower your default trust in that specific source going forward, not just for the next paragraph.
A fluent, well-formatted, confident-sounding answer outside your expertise feels more credible precisely where you are least able to check it.
"You were wrong about X, so ignore everything you say" is not the disciplined recalibration this model describes; it is a different, unrelated argument.
the source has a documented, verifiable track record of accuracy specifically on the unfamiliar topic, such as audited financial statements or peer-reviewed primary literature.
the stakes are high enough that trust calibration alone is not enough. Check primary sources directly rather than estimating how much to discount a secondary one.
what you caught might be a reasonable simplification for a general audience rather than an actual factual error.
This browser tool structures your thinking. It does not send your information anywhere or perform AI analysis.
Save as Skill converts the model into portable Markdown instructions that teach an AI when and how to apply the Gell-Mann Amnesia Effect, what questions to ask, what evidence to seek, what mistakes to avoid and how to structure its analysis.
Save the skills you find useful and organize them into your own reusable library. The Markdown is intentionally model-neutral, so it can be used wherever an AI system accepts uploaded files, project knowledge, custom instructions or reusable text context. Platform capabilities differ, so this is a portable skill file rather than a claim of one-click installation.
No. It means your trust in a source should be calibrated by its demonstrated accuracy, not reset to a default every time the subject changes. A source that is careful in your field has earned more trust elsewhere than one that is careless in your field, even if you cannot check the rest directly.
"Don't trust the media" is a blanket stance that applies no matter what you actually observe. The Gell-Mann Amnesia Effect is a specific, personal signal: it only fires when you yourself caught a real, checkable error, and it asks you to carry that specific evidence forward rather than discard it at the page break.
It applies at least as strongly, and arguably more so. A single AI model answers questions across every subject in the same fluent, confident tone, so there is no change in writing style to cue you that you have left familiar territory. The visible calibration cue that a shift in publication, author or outlet might normally provide is largely absent.
Keep a running, specific memory of where a given source has been wrong, and let that memory follow you into every other topic from the same source. The discipline is not distrust by default. It is refusing to let an earned discount quietly expire.