How AI Slows You Down
There are loads of ways we’re becoming more productive with AI. Everyone’s talking about them. What no one’s really talking about is how it slows us down.
These are some of the pitfalls I’ve found in my own work, and I think they’re worth thinking about, both for individuals and for teams. None of this is anti-AI. It’s just a set of observations from my own day-to-day that doesn’t get said out loud.
1. I built myself a micromanager
Small tasks I used to just do, I now run through AI and then sit there reviewing. Take an email. It used to take me five minutes to write and I’d click send. Now I draft it, ask AI for a pass, get notes, go back, get more notes, and we go back and forth until the draft is “ready.”
I’ve basically hired a micromanager who gives me feedback on every tiny action. Emails, copy, captions, posts, marketing. I’m no longer just doing the task. I’ve introduced a workflow of create, review, revise, re-review, revise, and just like having a manager who won’t let you send an email without their review and approval, every tiny action now takes five times as long.
2. The context tax
When I don’t run that whole review loop and instead go straight in, I hit the opposite problem. I give it a quick prompt: “write me an email that says X and Y.” But it’s missing so much context that the output is generic, so now I’m going back and forth correcting it and feeding in what it didn’t know.
And if I do give it the full context up front, who I am, who they are, what’s going on, what the goal is, the background, all of it, that itself eats up so much time that, again, it would have been faster to just write the email myself.
Either way there’s a tax. Too little context and I pay on the back end fixing it. Too much context and I pay on the front end loading it in.
3. The “it’ll only take a few hours” trap
When something was high effort, the effort did me a favor. Three months of work forced a real calculation, a proper cost-benefit analysis, before I started. And most of the time that analysis said no, so I never began. The friction was doing my prioritizing for me.
Now that same task takes an afternoon, and I get pulled in by “it’ll only take a few hours.” So I do it. I’ll query all my clients to work out exactly how long they were on a call. What words they used the most. I’ll build a little tool. I’ll pull information that wasn’t accessible before. I’ll do some deep research on a topic. None of it ever cleared the old bar, but the new bar is so low that I never check.
That’s the trap. The old high-effort work was so obviously ROI-negative that I never started. The new low-effort work isn’t obviously ROI-negative, so I never run the calculation at all, because it feels almost free. But if I added up the time and energy, a lot of it goes into building tools and toys that aren’t that valuable, or that I never even end up using. We’re building a lot of cool reports and tools. Cool isn’t the metric.
4. The trap of 90%
This is the close cousin of the last one. One prompt, and suddenly you’re looking at 90% of the tool, the product, the result you wanted. It’s an overwhelming wow moment. It feels basically done.
Then the rest of the day disappears trying to turn that 90% into something actually workable and usable. The last 10% fights back, and it turns out to be most of the real work. And some projects never make it out at all. Prompting back and forth for hours, forever “allllmooost theeere,” close enough to keep me hooked and far enough that they never ship.
5. Keeping up with the tools I already use
The tools I already use are constantly shipping. New features, new models, new capabilities, new “this is the way to do it now.” Webinars, courses, documentation. A real chunk of my time goes to just keeping up with the tools I’ve already committed to.
6. The constant pull of new tools
Then there’s the wave of brand new tools. Something comes out, it looks great, and I want to try it. Sign up, get over the learning curve, figure out where it fits. New models, new companies, new approaches, all of it. Trying them out and staying on top of innovation takes a huge amount of time too and (other than filling my inbox with welcome emails) often leaves me with little to show for it in the end.
7. Watching everyone else push the tools further
Even once I’m settled into my own stack, I’m bombarded by amazing work people are doing with the exact tools I already have. There’s always someone on the cutting edge pushing my own tools to their limit and sharing it. Some AI SDR or marketing agent where, by commenting “Money while I sleep,” I get a fourteen-page PDF walkthrough on how they got a tool to do something I wasn’t doing before. And although clever and very cool, this often isn’t something I really need, and hardly ever justifies the tokens.
The last human job
I’m one of the biggest proponents of AI out there. I’m looking forward to being semi-bionic, treated by AI doctors, with AI friends and AI doing most of my work. This productivity loss is just what I’ve noticed in my own day to day.
It adds up to one uncomfortable idea: AI lowers the cost of doing things so much that we stop asking whether they’re worth doing. And I think that question is exactly where humans fit in an AI world. When AI can do anything and everything, the job left to us is choosing what’s worth doing.
