The Only Way Is Through

Part 2 of a two-part series on how to approach AI

In Part 1, I argued that the AI discourse is a prediction market full of narrow truths and structural incentives to dramatize them. Know who's writing and what they want to be true, then read accordingly.

The reasonable next question is: Ok, but then what?

The honest answer: you can't read your way to insight. The only way to build real intuition for what AI can and can't do is to use it.

The deluge of conflicting takes makes it feel safer to stay on the sidelines and adopt a posture of skepticism. But skepticism without engagement is avoidance, just with better branding.

The antidote is surprisingly simple; it comes down to three things:

Tinker: Put your hands on the tools

There is no substitute for simply using AI. Carve out dedicated time to explore. Sit down with any AI tool and start prompting. Ask it what would make a good prompt. That alone starts to reveal how the technology responds, where it breaks, and what it's actually good at. The gap between hysterical predictions and real capabilities becomes obvious fast once you have your hands on the keyboard.

Everyone tinkers differently, and that's how it should be. For me, it was a school email ingestion engine. My wife and I were drowning in school communications, which led to forgetting it was picture day or a half-day dismissal. I built an AI system to parse the dates and details and surface what matters. That single project taught me more about what AI can actually do than anything I'd read.

Find an AI-native: Recruit a tinkering partner

Find one person in your life who's already deep in this stuff and talk to them regularly. They don't need to be a thought leader or have a following. They just need to be someone in the weeds, tinkering themselves, and someone you actually enjoy talking to. That last part matters. If the conversations aren't fun, they won't sustain.

I have several, including Conrado Marturet and Luke McGartland, founders we’ve backed at Penny Jar and born tinkerers. Every conversation we have devolves into the intricacies of the latest models or some new tool they’re testing. Their fervor for the technology has compounded my own, and I look forward to every conversation.

Pursue Joy: Use AI for something you love

Find something you love and use AI to pursue it. Joy is the engine that sustains tinkering past the point where novelty wears off. The people who are furthest ahead right now aren't necessarily the smartest. They're the ones who found a use case that made them want to keep going.

For me, it was March Madness. My kids wanted to fill out a bracket, but the typical approach is meaningless to a preschooler. So I used AI to build a picker where they could choose winners based on each school's mascot, color scheme, logo, and location. They had a blast. I had a blast. Building those personal use cases is where real understanding takes root.

Get in the arena

Kevin Weil, OpenAI's Chief Product Officer, put it simply: "The AI models you're using today are the worst AI models you'll use for the rest of your life." I'd add that AI plays the smallest role in our lives it ever will. Every day it will play a bigger one. The question is whether we shape that role or let it shape us.

Skepticism is healthy. Paralysis is not.

Tinker. Find an AI-native. Pursue joy. Get in the arena.