Fluency, Friction, Finite
Three words I’m focused on this summer for my kids, my work, and myself.
Every few months I try to step back from the headlines and ask: what actually matters right now in the AI transition? Summer is busy and chaotic as routines get disrupted. Your attention may drift from the impact of AI on your work and your kids, but the pace of change won’t slow down. Here’s how I’m focusing between now and the start of the next school year.
Fluency
I’m focused on AI fluency, not literacy. Ohio State’s AI Fluency initiative uses a definition I think should become the objective at work and in higher ed: fluent in your field of study and fluent in the application of AI in that field.
That definition inextricably links human expertise and AI capability. You can’t be AI fluent without knowing your domain deeply enough to guide AI effectively. You also can’t be fully fluent in your field without understanding how AI is changing the work within it.
This is the standard I’m pushing in my role leading AI strategy at my company. It’s also how I think about what I want for my kids. I want them to develop mastery of the knowledge in their fields of interest AND the ability to critically assess how AI can be applied to advance those fields.
In my own work, AI fluency requires matching the right intelligence (human or machine) to what the task actually requires such as choosing the most capable frontier models for complex strategic work vs. lighter, faster models for routine tasks. That judgment to apply the right intelligence to the right task at the right moment is a fluency skill that will increasingly be a professional expectation.
The goal isn’t to become an AI developer or understand the nuances of every model. The field is moving too fast for that. The goal is to understand AI’s evolving role in your work, your field, or your learning so you can responsibly shape its use.
Friction
At this point in the AI transition, I think the primary risk of AI isn’t the technology itself. I see increasing risk in the frictionless use of AI, and frictionless use is exactly what the frontier AI companies are driving toward.
Frictionless AI puts pressure on humans to operate at machine speed. In doing so, we often bypass deeper thinking and productive struggle in an attempt to keep pace.
I can see the trends and the risks they create (especially but not exclusively for kids), so I’ve built something to counter this. It’s a free Chrome browser extension called f/riction that creates structured moments to help students think before AI thinks for them. The first version helps students pause and plan at the moment they reach for AI. The version I’ve been developing this spring goes further. It builds friction defaults into the AI session itself: instructions that change how the AI behaves during the interaction.
A student (or any user) can choose from five types of reflective friction:
Check my direction: ask clarifying questions so I continue to shape the work
Flag uncertainty: reveal what you’re unsure about so I can verify it
Push back: challenge my weak spots and assumptions without taking over
Ask for my reasoning: understand my thinking and ask questions that sharpen it
Highlight blind spots: point out missing perspectives I haven’t considered
The friction is designed to counter familiar AI failure modes: overly confident answers that discourage scrutiny, sycophancy that tells you what you want to hear, and the propulsive momentum of a conversation that obscures better alternatives.
You don’t need the extension to try this. The next time you open an AI tool, ask it to find weaknesses in your argument, be transparent about its confidence, or interview you before jumping to an answer. See how different the interaction feels when you build in even a small moment of friction.
Finite
AI use should have limits. Not because the technology is bad, but because we (our kids, our companies, ourselves) aren’t ready for limitless, frictionless AI use.
From the initial release of ChatGPT until recently, users have been trained that AI access is essentially limitless. Pay a monthly fee and run whatever you want. The frontier labs heavily subsidized use to promote adoption, painting a vision of intelligence that is “too cheap to meter.”
That assumption is already breaking down. Uber recently disclosed that it burned through its entire 2026 AI tool budget in four months after deploying AI coding tools to 5,000 engineers. Microsoft also pulled back internal access to AI coding tools citing unsustainable costs. Why would we think that any company has an infinite set of high-value ideas ready to implement that could justify unlimited AI use? The imposition of constraints (not abundance alone) has often been essential to unlocking more transformative innovation.
Finite isn’t just a corporate finance story and it isn’t anti-AI. It connects to an AI Waypoint I come back to often: preserve untouched space. In a world of constant algorithmic noise, protecting time for human-only thinking is essential. That can be as simple as a bike ride with the music off, an evening with no screens, or choosing to read a long document instead of asking for an AI summary.
The choice to impose limits on AI use has the added benefit of creating friction. When you can’t throw unlimited AI at every problem, you have to plan. You have to decide what’s worth the cost and what you’re better off doing yourself. Whether measured in tokens, dollars, or hours, the Finite AI use limit becomes a forcing function for the intentionality that Fluency and Friction are trying to build.
The summer window
The summer provides a critical window to step back, reset, and focus on the things that matter most. For the AI transition, Fluency, Friction, and Finite are a good place to start, and you don’t need to tackle all of them at once.
If your kid just finished a school year where AI use felt confusing, start with Fluency. Ask them what AI looked like in their classes this year or how AI might relate to something they are interested in. If you’ve been using AI at work and noticed your own thinking getting lazier, try adding Friction to your next session. If everything just feels like too much, start with Finite by protecting one space this week that stays human-only. But Finite doesn’t mean None. Keep exploring AI, keep learning, but also keep limits that force your brain to stay engaged.
If you want to go deeper on any of this, reach out. I’m happy to share what I’m building.
— Whit
