Hank Green’s AI confession: dopamine hits aren’t a substitute for real work

Hank Green, the guy who’s been making science and internet culture digestible for over a decade, just said something that a lot of us probably need to hear. In a recent video, he apologized for how much he’s been using large language models, calling the dopamine hit he gets from them “not healthy for me or good for the world.”

That’s a striking admission from someone who’s built a career on being curious and experimental. Green isn’t some Luddite or a tech skeptic. He’s the co-creator of Crash Course and SciShow, a guy who genuinely loves learning and sharing knowledge. So when he says his AI usage is messing with his head, I pay attention.

The dopamine angle is the part that really resonates. We all know the feeling: you type a prompt, the model spits back something clever, and you get a little jolt of satisfaction. It’s like a slot machine that occasionally pays out with a witty paragraph or a half-decent explanation. But Green’s point is that this isn’t the same as actually doing the work. It’s a shortcut that feels productive but might be making us less capable of sitting with a hard problem.

I’ve felt this myself, to be honest. There’s a temptation to use an LLM as a thinking crutch, to let it generate the first draft of an idea before I’ve even fully formed my own. It’s efficient, sure, but it’s also a way of avoiding the uncomfortable part of creation—the part where you’re not sure what you think yet. Green’s confession makes me wonder how many of us are doing the same thing without admitting it.

What’s also interesting is that he framed it as “not good for the world,” not just for himself. That’s a bigger claim. It suggests that our collective reliance on AI-generated content might be eroding something—originality, depth, maybe even our ability to surprise ourselves. When everyone’s using the same models, the outputs start to converge. You get a kind of bland, averaged-out creativity that’s safe but forgettable.

Green’s apology isn’t a call to abandon AI altogether. He’s not saying the tools are evil or that we should all go back to typewriters. He’s just being honest about the trade-offs. And that honesty is rare in a space where everyone’s either hyping the tech or doom-posting about it. He’s giving us a middle path: use the tools, but be aware of what they’re doing to you.

I’ve been thinking about my own usage since I watched his video. I’m not going to swear off LLMs—they’re genuinely useful for certain tasks, like summarizing research or brainstorming when I’m stuck. But I’m going to try to be more deliberate about when I reach for them. If I’m using ChatGPT to avoid thinking, that’s a problem. If I’m using it to augment my thinking, that’s fine.

The fact that Green felt the need to apologize at all is telling. It means he noticed the behavior in himself and decided it was worth addressing publicly. That takes guts, especially when your audience is full of people who might see AI as the future. But he’s always been good at modeling intellectual honesty, and this is a prime example.

I don’t know if this will change how anyone else uses LLMs, but it’s made me pause. The dopamine hit is real, but so is the cost. If the guy who’s been teaching us about the internet for years says it’s not healthy, maybe we should listen.