What Can't We Know About AI? | Knowing the Mind You're Working With | The Infinity Machine | and more...
Artificiality Summit 2026: What Can't We Know About AI? In the first of a series about our speakers
Remember how I told you how Dave knows I skip from A to D? I want to go back to that story and explain something important about how humans interact that is relevant for the way we use AI.
I explained how he's watched me do this for fifteen years. To make those logical leaps that feel obvious to me but are invisible to everyone else. Then when people ask me to backtrack and explain B and C, I get defensive. This is a bad reaction. I've worked on it forever.
Dave waits. He's learned to read the difference between "Helen is being difficult" and "Helen has seen something she can't articulate yet." He asks the right questions to pull B and C out of me. Sometimes he sees that there is no B and C—that I've made a leap that doesn't actually hold—and he tells me that too.
This knowing didn't happen quickly. It took years of working together, fighting, making up, watching each other succeed and fail. He's seen me under stress and relaxed, confident and uncertain, honest and self-deceiving. His understanding of how I think comes from how I've lived in front of him.
Psychologists call this Theory of Mind—the ability to understand that other people have beliefs, desires, and ways of thinking different from your own. Children develop it around age four, when they can pass the false-belief test: understanding that Sally will look for her marble where she left it, even though the child knows it's been moved.
Dave has fifteen years of Theory of Mind about me. He doesn't just know what I think. He knows how I think, where I get stuck, what I'm likely to miss, what I'm likely to rationalize. This has been essential for our work. It's also been essential for keeping me honest when my thinking goes wrong.
Now I collaborate with AI too. The AI builds a model of me—my vocabulary, my patterns, my preferences.
Do I understand how it works well enough to use it wisely? And do I understand what it's learning about me?
We reserve some of our writing for human eyes only, not machines. Subscriptions are how we keep it that way—AI crawlers can't reach what's behind the gate. A free subscription gets you access to everything. You can unsubscribe anytime and choose whether to receive emails in your member settings.
SubscribeAI is changing how you think. Get the ideas and research to keep you the author of your own mind.