The Cognitive Overload of AI Development
just one more prompt
Typically, when the Harvard Business Review publishes something, especially techy, people tend to pay attention. Well, that is, unless it goes against the ultra-psyops-capitalism that drives most of the known world, in the form of extracting every useful drop of blood and life from the glassy-eyed masses that are too exhausted or addicted to the doom-scroll to look up for a minute.
Hey, am I just the kettle calling the pot black, while I sell you a 10-dollar-a-month subscription? Maybe.
Just when you thought the late-night working, on-call fearing, Slack notification twitching, deadline anxiety-ridden Software Engineer was at the end of his or her rope, one JIRA ticket away from giving it all up for chicken farming … the world in all its maniacal plotting dropped AI square into the face of every developer.
Thanks to Delta for sponsoring this newsletter! I use Delta Lake daily, and I believe it represents the future of Data Engineering. Content like this would not be possible without their support. Check out their website below.
All we have experienced and learned to this point has been nothing, just training, really.
It was just preparation to be plugged into the actual matrix, the Gas Town crazed C-suite and CTOs, whose eyes shine bright with endless Claude-driven possibilities of Agents upon Agents, calling each other in endless token burning loops, while investors pour bags of money on their already rich heads.
Every normal path of the software engineering life cycle has been turned upside down, short-circuited, smashed, and thrown out the window.
In a blindly dizzy change of culture, the once clean-code-loving, JIRA-embracing, Agile acolytes have abandoned their first love for an Anthropic mistress, abandoning overnight 50 years’ worth of collective experience and knowledge, which was stripped away, sucked up, and fed into the darkness of Foundational Models as training data.
It wasn’t enough to simply turn a million programmers into token fools; once independent thinkers, we are all now sucking at the teat of subscription plans. We live in fear of being cut off from our new drug of choice, the token, the bringer of life and features.
The outcome and human cost of these Agentic Coding tools are as obvious as they come.
“We found that the phenomenon described in these posts—cognitive exhaustion from intensive oversight of AI agents—is both real and significant. We call it “AI brain fry,” which we define as mental fatigue from excessive use or oversight of AI tools beyond one’s cognitive capacity. Participants described a “buzzing” feeling or a mental fog with difficulty focusing, slower decision-making, and headaches. This AI-associated mental strain carries significant costs in the form of increased employee errors, decision fatigue, and intention to quit.”
- Harvard Business Review
What it means to program in the Age of AI.
I feel I might have some insight here, having been on the giving and receiving end of the AI firehose of code, and now that we’ve been living in that world for a bit, it’s pretty obvious both the upsides and downsides we are dealing with in this Brave New World.
Look, my personal feelings on it all are irrelevant, as are yours. We are just little cogs in a big machine, washed along the flood of professional life, victims of the rains of the tech culture at large. You may tell yourself you sit aloof, you and your neovim, but the truth is you’re doing what you’re doing because you’ve been influenced, or pushed places by necessity.
So what does it mean to deal with AI in all its reckless glory?
1. Senior+ Engineers have turned into glorified code reviewers.
2. It isn’t possible to keep up with the pace of AI output in a codebase.
3. You no longer understand the minutiae and details of an AI-generated codebase.
4. Your mind and soul are torn between doing the right thing (understanding and
reviewing changes in depth) and meeting expectations.
5. Inexperienced devs and non-engineering stakeholders think they are smarter
than they are (armed with Claude).
6. Bad designs and architecture are amplified.
7. You will get more “burned out” faster.
8. You let “things slide” that you normally would not have.
9. We have to deal with more organizational and professional chaos and uncertainty.I think what it boils down to, and the end of the day, is the mental burden of the expectation that we use AI to move quickly and produce more, while still being held accountable for all negative outcomes and side effects of the software products we produce.
It is a classic problem: being held responsible for something you can’t control in its entirety.
A little interwebs search will assure you that this topic is on the back of the minds of a lot of folk, who know enough to be worried.
Why kick against the goad?
So what?
It’s all good and well to rage against the AI machine, but methinks that might be a fool’s errand. There probably isn’t much hope to make any gains against the flow of tech culture; it’s probably best to just find our way through it, if we can.
Now, more than ever, it’s important to do a few things to counteract AI burnout in your own life.
Find hobbies and take time away from the computer.
Go outside, exercise, and read.
Treat AI coding like another skill that’s important in the marketplace, don’t overemphasize it.
Remember why you fell in love with coding, and do those things yourself regularly.
Ignore the AI Doomers and the AI Groomers at the same time.
Take the middle road.
I think a healthy dose of reality in all its forms is a great antivenom for a mind and body plagued by AI burnout. For me, this burnout creeps in slowly and takes over before I know it, or realize it.
Context switching quickly, massive code changes, and PRs, the incredibly fast pace of development and new features, and worrying about quality, understandability, and best practices.
Slow yourself down.
While those around you move and ship at breakneck speeds, you should slow down a little. Take time to think through large design decisions, architecture, and systems.
Think critically and slowly about that AI-generated PR, just because it was spewed out in a day (what once would have taken a week), spend an extra day or two really understanding the business context, what’s happening, and hidden decisions being introduced.
Beat them at their own game.
Don’t try to escape the game; it’s been forced upon you, and it’s probably not going anywhere. Be optimistic in your attitude, remember why you do what you do, and why you love it.
They think AI will make them a 10x Engineer with little work and no sacrifice. You know better.
Go deep, understand problems like no one else. Ship at their pace when needed, and slow down when it counts. Take your time. Find joy in the great outdoors, and live a healthy lifestyle. The code will flow long after you are gone, and was flowing long before you arrived. Remember that.
Please leave a comment and let me know how you view the AI revolution and how you are dealing with it yourself.
I feel I engage my brain much less than before, being constantly reviewing code a much more passive (but still exhausting) process than writing code. So I procrastinate anything that requires actually an engaged brain, like writing a design document.
Because it _feels_ more productive and addictive to just trigger 5 agents to write actual code even if it isn’t the actual business priority. And doesn’t need such an engaged brain. But at the end of the day, I don’t feel the satisfaction I used to feel pre-AI.
Thank you for writing this, I’ll focus on slowing down and enjoying what I do, even if it takes more brainpower.