Frenetic thinking

by Tom Johnson on Apr 26, 2026
categories: ai • writing

One of the things I dislike about AI is how it takes over everything — every conversation, every process, every tool. But this week I noticed it started taking over my thinking as well.

Here’s what happened. While building reference docs for an SDK I support, one of the APIs failed due to some conflicting visibility conditions. I was jumping the gun on writing the release notes before I had all the launch entry info and Gerrit previews, so I wasn’t exactly sure what the failed build was supposed to produce. But I wanted to work on these docs and get out the release notes early.

I immediately set AI to troubleshoot the failed build, and it highlighted the conflicting visibility conditions. I jumped from solution A to solution B to solution C, leaving a trail of bug comments and chat threads that were frenetic and hasty. It wasn’t my usual measured, careful thought process. I blame AI for this sped-up mindset. I seemed to lack the patience to fully troubleshoot the issue in a more careful, thorough way before recommending the solutions.

In retrospect, I should have waited until the launch entry info and Gerrit previews were available; then I could have corroborated the doc failures against the corrected tags that would lead to the desired outputs. But instead, I thought I could run blind without this info, and troubleshoot the issue, etc., and fix it all at blazing speed.

As I ended the day, I relayed my concerns to a German engineer. Instead of immediately jumping to a conclusion, he wrote out a lengthy doc analyzing and articulating the issue and its solutions. The doc was a clear example of careful, measured and even-handed thinking. Somehow in the immersion of AI tools, I’d forgotten this mental state. I’d forgotten the ability to resist jumping immediately into a chat thread or becoming impatient when someone doesn’t immediately respond to a message. What happened to the ability to carefully read a long document, or book? AI surrounds us with instant responses, instant analyses, and instant thinking.

Working with tools that can analyze entire code bases in a matter of seconds, and which can write out a lengthy bug description almost instantly, it’s hard to recapture slow mode. My brain is too impatient for anything that isn’t instant.

AI leads us to expect immediate responses for large, cognitively taxing tasks. This isn’t necessarily bad, but it does lead to situations like that one I described above, where I end up jumping from solution to solution in a frenetic, scattered way, quickly trying what works.

I see a lot of value in slow, deliberate thinking. Given that we work with instant tools all day, how can we still retain the benefits of slow thinking among so many instant responses?

Perhaps we can exert more scrutiny on the instant answers. We can implement more verification workflows to apply to those instant answers before accepting them.

Additionally, perhaps reading print-form books might be something that works for me, which I might emphasize as a deliberate practice in the slow movement to push back against instantaneous everything. Is a print book really the antidote to instant thinking?