Once a day, I now force myself to make an old-fashioned phone call.

Not a Zoom meeting. Not a Teams meeting. Not a recorded conversation with an AI notetaker quietly humming in the background.

Just a phone call.

I do it because afterward I have to reconstruct the conversation from memory. I have to remember what was said, identify the themes, connect the dots, and then summarize it for my AI agents. In other words, I make myself do the work first. The technology comes second.

It is a small habit, but one I have adopted because I am thinking a lot about the doom-and-hype cycle surrounding AI. Every conversation seems to swing between extremes. AI is either the second coming of the Industrial Revolution, unlocking unprecedented productivity and entirely new industries, or it is about to eliminate every job, drain our resources, and leave us hiding on remote ranches while robot towers patrol the horizon.

The truth, as it usually is, is somewhere in the middle.

Oddly enough, I have been thinking about all of this through the lens of a novel I just finished, Yesteryear.

The book follows a social-media “trad wife” who suddenly wakes up in 1855 and has to become a real one. What struck me was not the time travel, but the way the novel exposes our tendency toward extremes.

It presents two narratives for a woman’s life. Devote yourself entirely to your family and lose yourself in the process. Or climb the corporate ladder, answer to men in power, sacrifice everything for your career, and end up just as depleted.

As the CEO of a company and the mother of seven, I know firsthand that neither story has to be true.

Ironically, it was my 19-year-old daughter, Sofia, who urged me to read the book. We came away with the same conclusion: there has to be a middle path.

You can see her generation searching for one.

If my generation graduated hoping for internships at Merrill, McKinsey or Morgan Stanley, hers is starting businesses that pressure-wash garbage cans. The startup ideas they come up with are rarely glamorous. But that is the point. They are optimizing for autonomy, for control over their schedules, their work, and their lives. I sometimes wonder whether our obsession with extremes pushed them toward the middle.

A few weeks ago, Fortune published two essays from the two of us side by side. We wrote them independently and did not read each other’s until they were finished.

Mine was titled, “Three Times the CEO I Was a Year Ago.” Hers: “I Don’t Want AI to Think for Me.” She was deeply skeptical of AI. I was unabashedly optimistic.

In the weeks since our essays ran, I have found myself thinking less about what AI will do to the economy and more about what it might do to me.

What parts of my own thinking am I outsourcing?

What skills, painstakingly developed over decades, could quietly begin to atrophy?

That worry is not mine alone. Psychologists have long studied “cognitive offloading” — the way leaning on tools like GPS goes hand in hand with remembering less of the route ourselves. And in a 2025 study, researchers at Microsoft and Carnegie Mellon University surveyed knowledge workers and reported that the more people trusted generative AI’s output, the less critical-thinking effort they applied to checking it.

For me, the answer is listening.

Throughout my career, my greatest advantage has never been that I took the best notes. It was that I listened deeply enough to synthesize ideas in real time — to hear patterns, connect concepts, and ask the next question.

Today, every meeting can be recorded. Every conversation can be transcribed, summarized, analyzed, and critiqued by AI. That capability is no longer an add-on: AI notetakers ship as built-in features of the video-conferencing and office software companies already license, and a 2024 Microsoft-LinkedIn survey found that 75 percent of knowledge workers now use AI at work. The pull toward divided attention is structural, not a matter of individual discipline.

The temptation is to stop paying full attention because the machine is.

That is the habit I am trying to resist.

So once a day, I go analog.

I make a phone call that is not recorded. Then I force myself to reconstruct it afterward before I ever ask AI for help. I am the one who has to remember. I am the one who has to summarize. I am the one who has to explain the through-line.

Only then do I hand it over to my agents.

It is my way of exercising a cognitive muscle I do not want to lose.

Because the most human part of listening is not transcription. It is noticing. It is sensing hesitation. It is recognizing what was not said. It is connecting ideas before anyone else sees the pattern.

If we all stop participating because AI will “catch us up later,” who is actually left in the conversation?

That question is already moving beyond opinion essays: schools, universities, and employers are drafting their first rules about when recording, transcription, and AI assistance belong in a conversation at all.

Perhaps that is the middle path AI is asking us to find.

Not rejecting the technology.

Not surrendering to it.

But using it in ways that make our own thinking stronger instead of weaker.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.

This story was originally featured on Fortune.com