Skip to content

Method

How we know what we say

Check this site as you would any confident answer. What is the source? What was tested? What might we have read wrongly?

Fact

First-hand reports, not independent verification. Public cases use unnamed descriptions and director-reported labels. Private research keeps the original identities and dates. We do not add numbers they did not record.

Interpretation

Our reading of facts. People may disagree. We show our reasons.

Illustration

Examples that show the point. They are not reported cases.

Planned

Work we intend to do, such as recording review effort. Planned work is not completed work.

Our view and its limits

We regard common sense as innate human sense, not something a workflow creates. This is our view. It is not proof that everyone is born with fully formed common sense. People can judge wrongly. The exercises apply perception and understanding to decisions.

You can know how to judge work without knowing how to use production software. AI is trained on data and can reason. Fluent output and human certainty both need checking.

Research has its own limits

  • A 2024 PNAS study examined individual and collective common sense. Agreement varied across claims; it did not establish a universally shared standard.
  • Shen and colleagues, June 2024 found that cultural context and query language affected the LLMs they tested.
  • Bian and colleagues, May 2024 found that the tested ChatGPT could express relevant knowledge yet struggle to select and apply what a question needed.

These findings concern those studies’ tasks and tested models, not every model available today.

Attention to the particular

We question the starting point and look closely at the situation. The Stanford Encyclopedia’s Spring 2026 account of Aristotle’s ethics describes practical wisdom as acting well in a particular situation. It involves experience and thought, not just general rules. This is background. It does not prove that common sense is wholly innate or support this site’s AI conclusions.

We choose to look closely at a gesture, a pause or a change in meaning. No filmmaker endorsement or attributed quotation is claimed.

What we do not claim

  • That AI cannot reason, critique or propose criteria.
  • That human review is always correct. Humans also drift, tire and carry bias.
  • Any measured rate of errors, time saved or review cost.
  • Endorsements or testimonials of any kind.

What the Decision Gate checks

The tool helps you record your reasons. It checks that you named the people affected, criteria and reasons. It never calculates a verdict. Veto conditions restrict the outcome but do not choose it. Nothing is stored or sent. Everything runs in your browser.

Status

Editorial update: 13 November 2026. The owner’s later view replaces the earlier practice slogan. Original case dates remain unchanged. Free-paper edition 1.1 separates that view from research. It does not verify a past case.