AI Does Your Job. So Why Are You Still Performing It?

The person who looks busiest in your office is almost certainly not the one doing the most work. In 2026, when AI agents handle meeting notes, ship code, and build slide decks on demand, knowledge workers are still burning roughly half their energy on looking productive. A recent community poll made the rounds with a gut-punch result: 1,107 respondents said the most valuable skill at their job isn’t doing the work — it’s appearing to do the work. It reads like a joke. It isn’t.

The Theater Got Better Production Values

There’s a term for this: performative work. Optimizing the appearance of effort over the substance of it. Keeping your Slack dot green at all hours. Sending an email at 10:47 PM so the timestamp shows up. Workshopping a comment in advance so you can drop it in the standup and look engaged.

You’d think AI would kill this behavior. It hasn’t. It just upgraded the staging. The familiar scene now: an employee gets a report from Claude or ChatGPT in 30 minutes, then sits on it for two days before submitting. Scroll the replies on any “AI productivity” thread on Hacker News or Reddit and you’ll find the same confession over and over: if I turn things around too fast, they’ll just give me more.

AI Saved You Time. Where Did It Go?

If your tools made you twice as fast, you should have twice the breathing room. You don’t. Three reasons:

One, expectations scaled with the tools. A deck that took three days now takes an afternoon, and your manager has already updated their priors. Two, organizations are reflexively packing the saved hours with more work — efficiency becomes throughput, not slack. Three, and this is the real one: AI being able to do your job makes you nervous about your job. So you compensate with visible activity. If a model can do this in five minutes, why am I here? Performing work becomes the answer.

The Visibility Trap

Organizational psychologists call this visibility bias — managers default to evaluating activity because activity is legible, and output increasingly isn’t. Remote work made it worse. The market for keystroke loggers, mouse trackers, and periodic screen capture has exploded since the pandemic, and the surveillance tooling industry doesn’t seem to be slowing down.

Here’s the irony. Surveillance tools can’t measure real work, so employees learn to fake the proxies instead. Mouse jigglers, fake-typing scripts, and “active status” hacks pull thousands of stars on GitHub. The lesson is old but worth repeating: when a metric becomes a target, it stops being a metric. The more you measure activity, the more activity becomes the thing being optimized — not the work underneath it.

What Those 1,107 Votes Actually Mean

Back to the poll. 1,107 people agreeing that looking productive matters more than being productive isn’t a story about lazy employees. It’s a story about organizations losing the ability to tell the difference.

AI flattens output quality across performers. Two people submit the same memo: one prompted a model for five minutes, the other actually analyzed the data. From the artifact alone, you can’t tell. So managers retreat to the only signal they trust — who seemed to be grinding. That’s the failure mode. The measurement system broke, and the workplace adapted by getting better at theater.

The Fix Isn’t Personal Virtue

This isn’t a character problem. It’s an incentive problem. As long as the system rewards the appearance of work, rational people will produce the appearance of work.

Organizations need to evaluate what got built — decision quality, downstream impact, whether someone’s work made their teammates better — instead of who was online longest. Easy to say, hard to do. It requires managers who can actually read the work, which is the skill that’s getting rarer fastest.

For individuals, the honest question is uncomfortable: what are you doing with the hours AI gave back? If the answer is “stretching the work to fill them,” you’ve quietly joined the cast.

The scariest part isn’t the wasted hours. It’s getting comfortable in the role. The feeling of real focus, of pride in something you actually made — those atrophy fast when you stop using them. In the AI era, the thing most worth protecting may not be your job. It might be your sense that the work means something.

So here’s the question worth sitting with tonight: of everything you did today, what fraction was built, and what fraction was performed?