aislop.day
THURSDAY, 16 MARCH 2023

Incident File for Tone Slider Dependence

When asking an AI to make a message warmer becomes easier than trusting the first draft, the habit files itself.

3 MIN READai tool guilt

Incident Summary

At some point between drafting a message and sending it, the subject opened an AI tone tool, adjusted a slider from neutral toward warmer, and sent the revised version. The original draft was discarded. No record of this step was shared with the recipient.

This incident has occurred more than once.

Classification

The subject's behavior meets the threshold for procedural status under the governing rule: any case of tone slider dependence repeated twice counts as policy until someone deliberately breaks it. The subject has not broken it.

Assisted work becomes routine while disclosure stays exceptional. This is the mechanism by which a private correction becomes part of the operating workflow without any meeting to confirm it.

Material Findings

Tone slider dependence adds one private manual check. The check itself is brief, typically a few seconds, but it now precedes most external messages above a certain stakes threshold. That check has become part of the normal workflow.

Coworkers receiving messages produced by this workflow cannot detect it. The output is indistinguishable from unaided prose except in cases where the original tone and the adjusted tone diverge substantially, which the tool is designed to prevent.

The help is useful enough to use and awkward enough to disguise. Neither of these conditions is expected to change.

Contributing Factors

The conditions producing this incident are structural. An AI tone tool is available. The subject's first drafts are competent but not always confident. The gap between the two is exactly the width the slider was built to fill. This is not a failure of the subject. It is the predicted use case.

Note on Disclosure

No disclosure framework currently governs this category of assistance. The subject is not in violation of any stated policy. The subject is, however, operating in a gap that existing guidance has not reached, and doing so without acknowledgment. Routine use and continued privacy are both entirely rational responses to that gap.

Disposition

This incident is classified as procedural. No corrective action is recommended. The subject is advised that the private check now part of their workflow may become visible before guidance arrives to cover it.

TAGSai tool guilt
Share this