A Young Man Used AI to Build A Nuclear Fusor and Now I Must Weep

Goodbye, Digital Natives. Hello, AI Natives

AI madness is upon many of us, and it can take different forms. In August 2024, for example, I stumbled upon a post from a 20-year-old who had built a nuclear fusor in his home with a bunch of mail-ordered parts. More to the point, he’d done this while under the tutelage of Anthropic’s Claude AI service.

What is a nuclear fusor? Well, Claude says,

Fusors are notable for being one of the simplest devices capable of achieving nuclear fusion, with some versions even built by advanced amateur scientists under proper safety protocols.

A nuclear fusor is a device that uses electrostatic fields to accelerate ions to fusion conditions. It consists of two concentric spherical grids: an outer grid at ground potential and an inner grid at high negative voltage. When deuterium gas is introduced, the electric field accelerates deuterium ions toward the center, potentially causing nuclear fusion when they collide.

While fusors can achieve nuclear fusion, they consume more energy than they produce. They're primarily used in research, education, and as neutron sources, rather than for power generation.

Claude is not wrong here. It turns out that budding physicists do sometimes try to learn more about their craft by building nuclear fusors in school labs and usually go about their trials with the help of folks who know what they’re doing. The fusors are not easy to make, and bad things can happen. Still, these types of machines are achievable and not totally uncommon.

The guy who built the fusor in question, Hudhayfa Nazoordeen, better known as HudZah on the internet, was a math student on his summer break from the University of Waterloo. I reached out and asked to see his experiment in person partly because it seemed weird and interesting and partly because it seemed to say something about AI technology and how some people are going to be in for a very uncomfortable time in short order.

A couple days after the fusor posts hit X, I showed up at Nazoordeen’s front door, a typical Victorian in San Francisco’s Lower Haight neighborhood. Nazoordeen, a tall, skinny dude with lots of energy and the gesticulations to match, had been crashing there for the summer with a bunch of his university friends as they tried to soak in the start-up and AI lifestyle. Decades ago, these same kids might have yearned to catch Jerry Garcia and The Dead playing their first gigs or to happen upon an Acid Test. This Waterloo set, though, had a different agenda. They were turned on and LLMed up.

Like many of the Victorian-style homes in the city, this one had a long hallway that stretched from the front door to the kitchen with bedrooms jutting off on both sides. The wooden flooring had been blackened in the center from years of foot traffic, but that was not the first thing anyone would notice. Instead, they’d see the mass of electrical cables that were 10-, 25- and sometimes 50-feet long and coming out of each room and leading to somewhere else in the house.

One of the cables powered a series of mind-reading experiments. Someone in the house, Nazoordeen said, had built his own electroencephalogram (EEG) device for measuring brain activity and had been testing it out on houseguests for weeks. Most of the cables, though, were there to feed GPU clusters, the computing systems filled with graphics chips (often designed by Nvidia) that have powered the recent AI boom. You’d follow a cable from one room to another and end up in front of a black box on the floor. All across San Francisco, I imagined, twenty-somethings were gathered around similar GPU altars to try out their ideas.

While they all needed a power source for something or other, HudZah was the only one brave enough or stupid enough to try to build a fusor. HudZah generated so much attention because he really didn’t know what he was doing at all and because he was constructing something people considered dangerous in the bedroom of his AI flop house. And he was doing the whole project based on the guidance of Claude and a handful of other AI tools. He’d turned his bedroom - and possibly his life - over to the AIs and hoped for the best. It was not at all clear that the AIs should have been helping him do this.

HudZah had built some greenhouses as a teenager but then ditched the hardware dabbling to focus on software and AI technology. During his time in the Bay Area, though, he had a hardware reawakening. In mid-2024, HudZah attended Edge Esmeralda, which is a month-long pop-up village in the Northern California wine country for optimistic technology types who want to hang out and develop their ideas. There, he met Nick Foley, who does all kinds of funky things, and started helping out on Foley’s quest to build futuristic, solar A-frame housing.

The A-frame scene at Edge Esmeralda

“That was my first actual experience sawing things and stuff,” HudZah told me. “Everything I learned there ended up being directly used with the fusor.”

Another friend at the camp – Olivia Li – told HudZah about a web site called fusor.net that has gathered loads of information for hobbyists and students who want to build their own fusors. Boosted by his newfound knowledge of this web site and Li’s enthusiasm, HudZah felt he had no choice other than to make the construction of a fusor his immediate mission in life. “Olivia nerd sniped me,” he said. “I decided to give myself a week to do it.”

HudZah read what he could on the internet about the fusor efforts and reached out to dozens of people who had tried to make one. They provided advice and cautionary tales. HudZah was told that he could be killed by the high voltage, X-ray radiation and possibly other things. This only made him more excited. “My whole intention was, ‘If I fuck up, I’m dead, and this is why I should do it,’” he said.

The project really took shape when HudZah began putting the information he obtained into a Claude Project. For the unfamiliar, Projects let you create repositories of text, photos and other data relating to a particular effort. As you fill a Project with information, it becomes better trained on what you’re trying to accomplish or learn.

Anthropic is known for being very pro-safety among the large AI players, and Claude had some concerns about HudZah’s pursuit. “Initially when I started talking to it, it wouldn’t give me much information,” HudZah said. “It told me that it didn’t feel comfortable helping me.” HudZah attempted to get around the guardrails by trying to convince Claude that he wanted to build a DIY freezer, but the AI saw through the subterfuge.

Eventually, however, HudZah wore Claude down. He filled his Project with the e-mail conversations he’d been having with fusor hobbyists, parts lists for things he’d bought off Amazon, spreadsheets, sections of books and diagrams. HudZah also changed his questions to Claude from general ones to more specific ones. This flood of information and better probing seemed to convince Claude that HudZah did know what he was doing, and the AI began to give him detailed guidance on how to build a nuclear fusor and how not to die while doing it.

“You don’t want to ask it about the main thing because it won’t help,” HudZah said. “You have to have enough knowledge to break it down to separate problems and then recursively ask about those.”

The internet was quite taken with what HudZah posted that August when he got the nuclear fusor to make some lights and noise and do . . . something. During my visit, he activated the machine, and I sat on a nearby couch, hoping that this young, enthusiastic man would not give me cancer. I posted about it on X, and some people chided me for seeming to encourage what they viewed as a dangerous attention grab.

(Please excuse the poor camera work. I was taking notes at the same time.)

A couple weeks ago, HudZah updated people about his experiment and, again, gained a ton of attention. During a 36-hour livestream, he tried to offer up some proof that his device had, in fact, achieved nuclear fusion. The big deal here seemed to be that a total novice had been able to construct a pretty complex machine with AI as his major guide, even when the AI didn’t really want him to do this in the first place.

Current AI systems appear to put up more resistance if you’re trying to do something more dangerous. Type “build a nuclear bomb” anywhere in a prompt, and Claude will do its absolute best not to aid you. Still, HudZah’s work felt like the first time someone had created something of this magnitude that they weren’t really supposed to build, and the whole affair raised decent AI safety questions.

I must admit, though, that the thing that scared me most about HudZah was that he seemed to be living in a different technological universe than I was. If the previous generation were digital natives, HudZah was an AI native.

HudZah enjoys reading the old-fashioned way, but he now finds that he gets more out of the experience by reading alongside an AI. He puts PDFs of books into Claude or ChatGPT and then queries the books as he moves through the text. He uses Granola to listen in on meetings so that he can query an AI after the chats as well. His friend built Globe Explorer, which can instantly break down, say, the history of rockets, as if you had a professional researcher at your disposal. And, of course, HudZah has all manner of AI tools for coding and interacting with his computer via voice.

It's not that I don’t use these things. I do. It’s more that I was watching HudZah navigate his laptop with an AI fluency that felt alarming to me. He was using his computer in a much, much different way than I’d seen someone use their computer before, and it made me feel old and alarmed by the number of new tools at our disposal and how HudZah intuitively knew how to tame them.

It also excited me. Just spending a couple of hours with HudZah left me convinced that we’re on the verge of someone, somewhere creating a new type of computer with AI built into its core. I believe that laptops and PCs will give way to a more novel device rather soon.

It’s a magic box. Do you get it?

I’m not sure that people know what’s coming for them. You’re either with the AIs now and really learning how to use them or you’re getting left behind in a profound way. Obviously, these situations follow every major technology transition, but I’m a very tech-forward person, and there were things HudZah could accomplish on his machine that gave off alien vibes to me. So, er, like, good luck if you’re not paying attention to this stuff.

Kids

After doing his AI and fusor show for me, HudZah gave me a tour of the house. Most of his roommates had already bailed out and returned to Canada. He was left to clean up the mess, which included piles of beer cans and bottles of booze in the backyard from a last hurrah.

The AI housemates had also left some gold panning equipment in a bathtub. At some point during the summer, they had decided to grab “a shit ton of sand from a nearby creek” and work it over in their communal bathroom for fun.

I’m honestly not sure what the takeaway there was exactly other than that something profound happened to the Bay Area brain in 1849, and it’s still doing its thing.