Walking the Dog vs. Sitting in Meetings: Where Real Engineering Happens

Someone recently posted an article about how they see engineering changing with the advent of AI. They postulated that engineers will spend more time in group settings with end users and stakeholders to build solutions.

It really seems like engineering roles are expected to take on the role of Product Owners. Using AI to prototype solutions directly with end users seems logical; it reminds me of the JAD-style sessions I was involved in back in the 90s (ouch, that really dates me!).

However, I do not believe this will be beneficial beyond initial prototyping and MVP development. Why? It ignores the human element.

Let me explain this in two parts.

First: AI often misses the "Why." I’ve seen this repeatedly: the solution AI proposes for a particular problem is technically plausible but fundamentally wrong. As an example, AI was given a bug report stating "the upload is taking too long." AI analyzed the code and pinpointed specific functions to optimize. After deeper investigation, however, this proved to be incorrect. While there was room for performance improvements, the real issue was user perception.

The user assumed the upload was lagging based on the UI. Under the hood, the UI wasn't correctly reflecting that the upload had already succeeded; it was actually waiting on a separate process unrelated to the transfer. It takes a human to deduce that (which explains why good prompting is so critical for successful AI interactions). While making those assessments could be achieved in a social setting, I am confident that doing so by myself was faster and a better use of everyone’s time.

Second: The power of the "Diffuse Mode." That leads to my second point. I do my best analysis and thinking when I am not in a social situation. In fact, it often happens when I’m not even at a computer—walking the dog, at the gym, or driving to the store.

Neuroscientists call this the Default Mode Network (DMN), which activates when you aren't consciously focused on a task. This is the space of imagination, reflection, and integration. When the DMN is allowed to function freely, you make new connections, revisit unresolved questions, and entertain possibilities that don't emerge during focused work.

I hear a lot of push for engineers to work more in groups. I notice this comes mostly from people whose roles are inherently social—executives and sales roles come to mind. This makes sense from their perspective because they must be social to complete their jobs. But that does not necessarily translate into effective problem-solving for engineering.

While AI is a powerful tool for real-time collaboration and prototyping, it is a mistake to assume it will—or should—replace the solitary deep-thinking sessions where the most complex engineering problems are actually solved.