Honesty
Polish can make an unverified claim look sourced.
By the end of this lesson
- Separate fact from interpretation, prediction and opinion in your own work
- Disclose where a piece of work was substantially machine-assisted
- Verify the claims that carry consequences before they travel
Watch
Episode V07
The video for this lesson has not been produced yet. The narration below is the same script, and it plays now.
Transcript
We have entered an age where almost anything can look real. Which means trust has become more valuable than ever.
Honesty begins with a simple question: Am I being transparent?
AI can help verify consequential claims. Compare documents. Find inconsistencies. Separate fact from interpretation. And help us investigate information faster than ever before.
But AI can also make uncertainty sound certain. Incomplete research sound finished. And borrowed intelligence sound like personal expertise. That is where honesty becomes essential.
Before sharing something consequential: Verify the facts. Identify the source. Separate evidence from opinion. And when AI substantially helped create the work—say so.
Transparency does not diminish your work. It strengthens it. Credibility is not built by pretending we did everything alone. It is built by giving people a reason to trust the process.
Take one important claim you plan to share. Trace it back to the evidence. Then ask: Would I still say this if everyone could see how I reached the conclusion?
In an age of artificial intelligence, honesty becomes infrastructure for human trust.
The lesson
A generated paragraph can carry every surface marker of being well sourced — the measured tone, the specific-sounding figure, the confident attribution — without any of it having been checked. The markers used to be weak evidence of diligence, because producing them took some. They are not evidence of anything now.
The most useful habit here is mechanical rather than moral. Go through a piece of work and mark each substantive sentence as one of four things: something you know, something you are inferring, something you are predicting, or something you think. Most drafts turn out to contain far more of the last three than they appear to, because the register is identical across all four.
The second habit is disclosure, and the argument for it is practical. If a piece of work is substantially machine-assisted and that becomes apparent later, everything else you have produced gets re-examined. Saying so up front costs almost nothing and protects the rest of your record.
Verification is not required uniformly. It is required in proportion to consequence. A number in a note to yourself can stay soft; the same number in something a council will act on cannot.
Straight from the framework
Virtue 7
Am I being transparent?
- What it contributes
- Can make anything sound sourced.
- Where it goes wrong
- Technological polish disguising a claim nobody verified.
- The practice
- Separate fact from interpretation, prediction, and opinion.
Do this in the app
Put the lesson down somewhere real
Take one worksheet answer and mark each sentence as known, inferred, predicted or opinion.
Done when — You have found at least one sentence that was not what it looked like.
/dashboard
Go further
Check yourself