The Dopamine of Artificial Intelligence: Why LLMs Are Making Us Hooked to Our Computers

How I Built Three Websites Without Knowing How to Code—and Why I Couldn't Stop

I am not a computer scientist. I have never taken a programming course, never studied data structures, and I have never written a line of JavaScript in my life, at least not knowingly. Still, over the past few months, I built three fully functional websites and one working portal, all with the help of a Large Language Model.

It started simply. I opened the AI assistant with just one question: how does a website actually work? Two hours later, I was still at my desk, refining prompts, asking follow-up questions, and watching code appear on my screen that I could copy, paste, and run. A page rendered. A button worked. A form was submitted. Something in my brain lit up.

That something, I have come to believe, was dopamine.

This blog is my attempt to understand the strange pull I felt, and that millions of AI users now feel, by looking at it through neuroscience, psychology, and behavioural economics. Why does one answer lead to ten more questions? Why do we feel energised, not exhausted, after hours with an AI? And is artificial intelligence creating a new kind of dopamine loop, one based not on likes and notifications, but on insight, surprise, and intellectual reward?

What Dopamine Really Is (and Is not)

Popular culture calls dopamine the "pleasure chemical." Neuroscience, however, tells a more nuanced story. Dopamine is less about pleasure itself and more about anticipation, motivation, and learning. It is the chemistry of wanting, seeking, and moving forward.

The key concept is reward prediction error, a term from the work of neuroscientist Wolfram Schultz. Dopamine neurons fire most strongly not when we get an expected reward, but when we get an unexpected one—when reality turns out better than the brain predicted. The brain uses this surprise signal to learn: pay attention here, something valuable is happening, do this again.

Now think about what happens when someone with no computer science background types "build me a homepage for my organisation" into an AI tool, expecting confusion, but instead receives working code, a clear explanation, and an invitation to ask more. The gap between expectation and outcome is huge. The surprise is real. And in the brain, surprise is a kind of reward.

I should be careful here: we cannot say that AI "directly causes dopamine release" in any clinically measured way. No one has put prompt-writers in a brain scanner at scale. But the mechanism is plausible and aligns with what we know: novelty, anticipation, uncertainty, and unexpected success are precisely the conditions that activate the brain's reward system.

Why AI Is Different from Google

We have lived with computers for decades. Why is this happening now?

Earlier digital tools were for execution or retrieval. Google gives you links, but you still have to do the work of synthesis. Social media provides emotional stimulation such as outrage, envy, or amusement. Video platforms give you entertainment that is passive and pre-made. None of them actually converses with you.

LLMs are tools for discovery. They respond in a personal, adaptive, and creative way. Instead of just retrieving information, they create new combinations of it, shaped to your exact question. Each prompt can give you a new idea, a better phrase, a working code snippet, a business model, a metaphor, or a comparison you had never thought of.

When I asked the AI to help me build my first website, it didn't give me a tutorial meant for someone else. It wrote code for my site, explained my mistakes, and adjusted to my level of understanding. When I admitted I didn't know what "deploying" meant, it explained it like a patient teacher. This is interactive discovery, and it is a very different experience from search.

The AI Dopamine Loop

Here is the cycle as I lived it:

Curiosity leads to a prompt, which leads to a response, then surprise, then insight, and then even more curiosity and more prompting.

I asked how to make a webpage. The answer surprised me — it was achievable. That insight sparked a new question: could I add a contact form? Yes. Could I make it look professional? Yes. Could I build an entire portal where users log in and submit applications? Astonishingly, the loop feeds itself. The more you ask, the more the AI shows you. The more it reveals, the more possibilities you imagine. As more possibilities appear, it gets harder to stop. Three websites turned into a portal. The portal led to ideas for two more projects. Each success made me wonder: what else can I do that I thought I could not?

This loop is especially powerful for people who are intellectually curious, like writers, researchers, students, entrepreneurs, civil servants, and designers. The same trait that makes them good at their work also makes them more likely to get drawn in.

Surprise as the Fuel

Here is the key point: the most interesting thing about AI is not how correct it is, but how unpredictable it can be.

Sometimes the AI writes a sentence better than you could have. Sometimes it connects two fields you never thought to connect. Sometimes, after five failed attempts to fix a broken login page, it suddenly gives you the solution, and the excitement of that moment is unmistakable.

Behavioural scientists call this a variable reward schedule. Rewards that come at unpredictable times are much more likely to shape our behaviour than those that arrive on a set schedule. This is the same pattern that makes slot machines so compelling.

However, we should not take the comparison too far. AI is not gambling. The gambler's reward is empty, but the AI user's reward is often something real and useful, like a website, an essay, a policy brief, or a solved bug. At most, AI is like an intellectual slot machine with genuinely productive results. Still, it uses the same psychological triggers: uncertainty, anticipation, and occasional rewards. We should be honest about that.

Information as Reward

There is a deeper layer. Research on curiosity — including work by George Loewenstein on the "information gap" and by neuroscientists such as Celeste Kidd and Benjamin Hayden — suggests the brain treats information itself as a reward, much like food or money. We are wired to seek knowledge that reduces uncertainty.

AI gives us a lot of information very quickly. What once took reading books, searching papers, consulting experts, or days of slow thinking now happens in minutes. Every interaction might contain a hidden gem. AI makes searching for information a rewarding experience, and our brains, which are wired to seek information, respond to that.

Flow: When the Machine Becomes a Thinking Partner

There is a name for what I felt during those long building sessions: flow, the state described by psychologist Mihaly Csikszentmihalyi, in which a person becomes completely absorbed when challenge and skill are balanced, feedback is immediate, and progress is visible.

AI is almost like a machine for creating flow. The feedback is instant: a writer sees a paragraph improve in seconds, a coder sees a bug explained, a student sees a concept simplified, and an official sees a draft note take shape. For me, every prompt led to visible progress on a real product. The gap between thought and output disappeared. Hours passed. I didn't feel tired; I felt alive.

I think this is why so many AI users report the same paradox: "I spent six hours at the screen, and I don't feel drained." Flow is rewarding in itself, and AI provides the conditions for it whenever you want.

The Bright Side: Democratisation as Cognitive Amplification

Let me be clear: this is not an anti-AI essay. What happened to me is, in the larger picture, extraordinary and good.

Someone with no computer science training built four working digital products. The knowledge that used to be locked behind degrees, boot camps, and years of practice has truly been opened up to everyone. AI didn't just give me information; it gave me the ability to do things. It turned me from a passive consumer of technology into an active creator.

When used well, this dopamine loop helps people learn faster, think more deeply, try out ideas, write better, and get past creative blocks. Dopamine itself is not the enemy; it is the chemistry behind learning, exploration, and motivation. Without it, nothing new would ever be tried.

The real question is not whether AI engages our reward system. It is whether our use of it is purposeful or compulsive.

The Dark Side: Infinite Curiosity Without Completion

Here is the risk I noticed in myself. After the third website, I found myself opening the AI tool not to build, but just to wander. I compared frameworks I would not use, redesigned pages that were already fine, and asked "what if" questions without a clear decision in mind.

AI can lead to endless exploration without actually getting things done. It can cause prompt addiction, research loops, decision paralysis, the illusion of productivity, and a slow loss of patience for slow, independent thinking. The screen always offers one more door, and every door is truly interesting.

The danger is not that AI makes us think. The danger is that it may make us feel we are working even when we are only exploring.

The New Human Discipline: Stopping Rules

Because AI can generate infinite answers, the scarce skill of the AI age is knowing when to stop asking and start acting. A few rules I now try to follow:

Define the purpose before opening the tool. "I am here to fix the registration form," not "let us see what happens."
Limit prompt iterations. If ten refinements have not produced it, step away and think.
Keep exploration time separate from execution time. Wandering is valuable, but schedule it instead of drifting into it.
Ask after every answer: "What decision will I now take?" If there is no decision, there is no progress.
Convert every session into an output. A page shipped, a draft saved, a choice made.
Protect time for slow, unaided thinking. The brain must remember how to be its own engine.

Discipline in the AI age will mean not only asking better questions, but knowing when to stop asking.

Conclusion: Curiosity Machines and Sovereign Attention

AI may be the most powerful curiosity machine humanity has ever built. It speaks to one of our deepest drives: the desire to know what is behind the next door. I have felt that drive take me from total ignorance to four working digital products, and I am grateful for it.

But the same loop that helped me build my websites could just as easily have taken up months in endless, unfinished exploration. The technology will not set that boundary for us. We have to do it ourselves.

The future will not belong to those who spend the most hours with AI. It will belong to those who turn AI-generated insight into real-world action, and who treat the excitement of discovery as fuel, not the final goal.

AI gives us endless doors to choose from. Wisdom is knowing which door to open, which room to build, and when to step away from the screen.