I had an interesting and eye-opening experience the other day. It caused me to reflect on where I’m going, and where I think many of us are headed.

While I’m a strong proponent of using AI as a writing and thinking assistant, I physically typed every word of this story to share a personal experience and sincere reminder.

The past six years were spent as a startup CEO and author of a new book, and 20 years before that as a tech executive. People who know and work with me say I’m a strong communicator with a generally positive outlook (like the happy Canadian I am).

I recently had a meeting with someone and increasingly found myself getting frustrated. I’ve known and worked with this person for a long time. About an hour into the meeting, I was close to calling it quits on our project. What was that about?

Luckily for me, as a student of mindfulness, I was able to pull back from the conversation in the moment and start to understand why I was frustrated. I couldn’t fully analyze it in real-time, but stepping back provided clarity that my reaction was unnecessary and unhelpful. The rest of the meeting went well and we agreed on a joint plan of action. A much better result than if I had quit the project.

Post-mortem aha

Afterward, something dawned on me. During the past nine months, since exiting my previous company, I’ve been working very closely with AI. In the beginning, I was using AI for basic question answering (prompting) and research like most people do––perhaps a few hours a day. But after I had published my new book on agentic AI and began spinning up an associated consulting company (The River Group), things changed.

To help me with TRG, I had spun up a team of agents, including three sales agents focused on different segments, a marketing agent, a strategy expert, and a Chief of Staff to help manage the whole thing. Over the last two months, the process of creating and then engaging with this virtual ‘team’ meant that I was spending at least six hours a day, sometimes more, talking with agents.

And I do literally mean talking. About six months ago I basically abandoned the keyboard and began using a popular speech-to-text app to dictate instead of typing. The result has been over half a million words spoken in the last six months, supposedly in the top 0.2% of all users of this app. This resulted not only in faster text generation, but also a change in the nature of the text I created––shifting my mindset to more of a ‘conversation’ mode vs. think-then-type mode.

By now you have probably guessed the punchline. I went from six hours of talking to real people every day, to six hours talking to simulated ‘people’. And they aren’t the same thing at all. I realize now that this has had a real effect on me.

To be clear, I have specifically instructed my AI to not be sycophantic. Its prime directive is to push back, to question, and to not assume that what I say is correct just because I said it. And the corollary to that is I often play the devil’s advocate and make statements I don’t believe just to get reactions. My AI agents know this and they practice it themselves.

But I have become accustomed to one interaction pattern now that does not carry forward to human interactions. With the AI, I can take the pushback as long as I like but it ends when I eventually say, “Okay, that’s enough. Let’s move on.” Then, like a dutiful soldier, it moves on.

In the real world, smart, self-directed people don’t work that way. Which is why I was getting so frustrated after an hour of discussing a point that would have taken me five minutes to (insufficiently) resolve with my AI team members. I would have missed the frustrating hour of discussion that ultimately resulted in a much better answer than the one I initially had in mind.

Are people sometimes stubborn, unreasonable, illogical, and self-absorbed? Yes! And that difficulty is often what creates the friction that creates unexpected results.

Friction is what I had lost the feel for after months of spending more and more hours talking with AI instead of real people.‍

I’m certainly at the bleeding edge of this scenario, but I predict this issue will become increasingly prevalent unless we guard against it carefully.

Suggestions

• Track the hours ‘talking’ with AI vs. with real people. Don’t let it sneak up on you.
• Ask people you trust for signs that your communication style is changing.
• Don’t assume that directives you give AI to avoid being sycophantic will solve all these issues. There are many subtle ways that engaging with an AI is not the same as engaging with real people, even if the language sounds the same.
• Never forget that the inherent unpredictability of real people, including their capacity to annoy or even frustrate you, is a feature and not a bug.

The next few years will likely be an incredible journey for us personally, our businesses, and for society. It’s up to us how that journey shapes the world.‍

Help us help you and others

‍How are you co-creating the future in your business domain? Choose at least one action below to engage:

• Learn more about an executive workshop on December 3 in San Francisco
• Subscribe to the Creatorbase Brief weekly newsletter.
• Apply for membership with one of two plans. Choose wisely.