Why AI Made You Busier

The tool you adopted to save time quietly became the thing that fills it.

You brought AI in to free up your hours. A year later you have more tools, more tabs, more output -- and somehow less room than before. The thing that was supposed to clear your plate became the largest item on it.

This isn't a complaint about the technology. The model is genuinely fast. The paradox is real and it's yours: measurable activity up, the work that actually matters no closer to done. You're shipping more and finishing less of what counts.

The optimization is the new work

Look at where the hours actually go. Every week there's a new model, a new workflow, a sharper prompt framework, a tool that finally does the thing the last one couldn't. So you upgrade. You tune the setup, rebuild the pipeline, rewrite your prompts, compare the options. Tuning the system feels like progress, and in a narrow sense it is.

But tuning the system is not doing the work. It's getting ready to do the work -- and getting ready is the most comfortable place in the world to live, because it never asks you to find out if you're good enough at the actual thing. There's always one more optimization between you and the hard part. The setup is masterable; the hard part isn't. So you stay in the setup. It's an old irony with a new interface: automating the easy parts of a job tends to leave the human with the harder, more exhausting remainder -- the judgment, the exceptions, the deciding.

This is the more-research loop with a software budget. A trigger -- the ambiguous, exposing task. An automatic move -- improve your approach to it instead of attempting it. A payoff -- the clean satisfaction of a better system, plus the feeling of being on top of things. A cost -- the task, still untouched, and the afternoon. The loop just found the most legitimate-looking hiding place it's ever had, because "I'm building out my AI workflow" is something you can say out loud at a standup.

Busy is the symptom, not the problem

The reason this reads as busyness rather than avoidance is that the activity is real. Prompts sent, drafts generated, tools evaluated, tabs open. A day spent optimizing looks nothing like a day spent staring at the wall, even though it can accomplish exactly as much of the thing that matters. The tooling supplies an endless stream of legitimate-seeming things to do, and the loop is happy to do all of them, forever, as long as none of them is the move.

The move out

The uncomfortable test is one question: is the tool actually the bottleneck?

Almost always, it isn't. You already have a model good enough to draft the email, structure the doc, prototype the thing. The bottleneck is the decision, the conversation, the bet -- the part no workflow touches. When you catch yourself reaching for a new tool, ask whether the current one could already do the next step if you let it. If the honest answer is yes, then the upgrade isn't progress; it's the loop, wearing a subscription.

Stop optimizing your approach to the thing. The approach is fine. Do the thing.

Sources

• On the irony that automating the routine work leaves the human with the harder, more exhausting remainder -- monitoring, judgment, the exceptions: Bainbridge, Ironies of Automation, Automatica (1983).
• On the same pattern arriving with AI tools -- de-skilling, cognitive offloading, and the optimization burden shifting onto the user: De-skilling, Cognitive Offloading, and Misplaced Responsibilities: Potential Ironies of AI-Assisted Design (2025).