I do not use social media. I cut it out years ago because I did not want that kind of dopamine pollution in my life. But lately I have been asking myself: did I just trade one loop for another?

When I started using AI in my software engineering work, especially after agent mode became available, I thought: “This is amazing. I can write code faster, solve problems faster, be more productive, and still have brain energy left at the end of the day for personal life.”

I am definitely more productive at work. But the brain energy at the end of the day? Not so much.

I do not think I have an addictive personality (I was never tested, though), and I do not get easily hooked on alcohol or other substances.

But I am a type A personality who loves problem-solving, maybe too much (I am trying to work on that), and agentic AI has become a hard-to-stop habit for me.

The Dopamine Loop

You know that reward feeling when you do something well? Dopamine is part of that system, but not only as a “pleasure chemical.” It is also involved in motivation, learning, and reinforcement.

In the old days of coding, the ratio between challenge and reward felt much heavier on the challenge side. I spent more time trying to solve one challenge, and the reward came later, at the end of a longer process.

Now, in the age of AI, especially with agent mode, things move much faster. We have a problem, and even if I am not writing every line of code myself, I am still finishing a business problem. That gives me a strong sense of accomplishment. The cycle is much shorter now, so I can jump to the next challenge right away. At least that is true for me.

My interpretation is that these faster feedback loops reinforce the behavior. Each quick win nudges me to chase the next win.

That is why I ask if agentic AI is “like a drug.” I do not mean it as a clinical diagnosis. I mean it as a reinforcement loop that keeps pulling me toward the next challenge.

The Behavior I Noticed

I got to a point where I opened my computer on the train commute just to write prompts to the AI and see whether I could finish a new challenge before my workday ended.

Now, every time I solve a business problem, I get that reward feeling and immediately start running toward the next one. I am always thinking about the next challenge I need to solve.

This makes me more productive, but by the end of the day I feel more tired and with less brain energy.

I know a lot of this is on me. Slowing down, delivering, and still ending the day with energy is a choice. But my personality pushes me to deliver more, not less.

So I am a better worker in output terms, but with less fuel left in the tank each day. That is the tradeoff I did not see coming.

What I Am Afraid Of

If I am solving real problems, maybe that is not an issue. But I am afraid I might start creating problems just so I can solve them and tell myself I was productive. That is related to what I wrote about task snacking, solving easy or unnecessary things to feel busy instead of tackling what truly matters.

If you want to understand the habit side of this better, my notes on Atomic Habits cover how reward loops get reinforced over time.

Maybe it is a me problem, not an AI problem. But I suspect I am not the only one feeling this: the pull of the next challenge, the next dopamine hit, and the risk of burnout if we do not manage it well.

PS: Paradoxically, I used agentic AI to proofread and improve this post. I am using it again.

Sources

Photo by Steve Johnson on Unsplash.