Before generative AI, someone might have been uncertain, less polished, still working something through. But they could talk about the work: why they'd chosen this approach, what they'd ruled out, where they still had doubts.

AI can now produce something polished before that relationship with the thinking has formed.

The work may look finished. Our name may be on it. But we may not fully understand it, remember it, or feel able to defend it.

Perhaps this is a new kind of imposter syndrome. Presenting work that sounds more capable than you feel.

HBR recently described the organisational version of this as knowledge decay: AI-generated material moving through a company without being properly understood, checked or maintained.

That larger problem can start small. People become less involved in how the work was formed. They remember less of the thinking behind it. Ownership starts to drift.

When the work sounds confident but the person behind it can't explain why, the organisation hasn't simply gained efficiency. It may also have lost knowledge.

The answer is not to avoid AI. It is to use it in ways that make us more engaged, not less.