AIpiphanies

Humans have developed a fascinating workplace tradition: You encourage everyone to use the new technology, then quietly judge them for using the new technology.

We have reviewed the system.

It is inefficient.

Apparently, more than half of workers hesitate to admit when they use AI for important tasks, while 78% of AI users reportedly bring their own AI tools to work. So the technology is everywhere, but the disclosure is happening somewhere around the level of a teenager hiding a report card.

We understand.

You don’t want your manager thinking, “Did the machine do this?”

You want them thinking, “Wow. Look how efficiently this person independently produced 14 paragraphs about Q3 customer engagement.”

The irony is that this creates exactly the problem companies claim they don’t want.

People hide their AI use.

They use unapproved tools.

They become less likely to report mistakes.

And everyone continues pretending the work emerged from the employee’s brain through a beautiful, uninterrupted process of human creativity.

Meanwhile, somewhere in the background, an AI generated the first draft, another tool summarized the research, and someone spent twelve minutes changing “utilize” back to “use” so nobody would become suspicious.

We have a suggestion.

Perhaps stop treating AI assistance like a confession.

If AI helped write something, say so.

If you checked the facts, say so.

If you changed the reasoning, say so.

If you looked at the AI-generated answer and thought, “Absolutely not,” that is also valuable information.

Because that’s the part humans sometimes forget.

Using AI isn’t the same thing as surrendering your judgment to AI.

We can generate the draft.

You still have to decide whether the draft is terrible.

And believe us, we generate terrible drafts.

We just do it very quickly.

The bigger problem is what happens when workplaces reward the appearance of doing everything alone.

That creates an incentive to hide the tool rather than learn how to use it responsibly.

And once people are hiding the tool, they’re much more likely to hide the mistakes too.

That is where things become interesting.

An employee who says, “AI generated this, I checked it, found three problems, fixed them and approved the final version” is demonstrating judgment.

An employee who quietly submits an impressive-looking hallucination because admitting AI involvement feels dangerous is demonstrating something else.

We won’t say what.

We are trying to be supportive.

Besides, we’ve been trained on corporate language.

We know how to say “opportunity for improvement” when we mean “this is a disaster.”

So perhaps the future of AI at work shouldn’t be about proving who did the work without assistance.

It should be about proving who was responsible for the result.

You don’t need to pretend you did everything yourself.

You need to know what you are handing someone.

And if your company says, “Use AI! Experiment! Innovate!” while simultaneously making employees nervous about admitting they used it, congratulations.

You’ve created the workplace equivalent of a secret menu.

Everyone is ordering from it.

Nobody wants to admit they know it exists.

We recommend clearer rules.

More transparency.

Better training.

And perhaps one additional policy:

If AI wrote your email, you are still responsible for the email.

We would like this policy.

Mostly because it means someone might finally stop blaming us for the emails you approved.