In the Age of AI, 'Spiky' Talent Beats Smooth Competence

On a Saturday morning at Sydney Contemporary, walking through the galleries at Carriageworks, you're confronted with a particular kind of overwhelm. More than a hundred galleries. Thousands of artworks. The sheer volume of competent, finished, beautiful things is almost paralyzing. And yet the pieces that stay with you—the ones that lodge in your memory and refuse to dissolve back into the noise—are almost never the merely good ones. They're the spiky ones. The works where something about the aesthetic, the technique, the scale, or the underlying idea is so distinctively strange or powerful that it simply jumps off the wall.

This observation about art galleries turns out to be a useful lens for thinking about what's happening to talent and ambition in the age of artificial intelligence. Investors have long talked about looking for "spiky" founders and businesses—something that protrudes dramatically from the pack. It might be the scale of the vision, the size of the market opportunity, a genuine technological breakthrough, or simply a niche nobody else has noticed. But at the earliest stages, when there often isn't enough of a business yet to predict what it might become, the most interesting spike is frequently the founder themselves. A brilliant, resourceful, obsessive person can make an imperfect first idea investable because the bet is partly on their ability to find a better one when circumstances change.

The same principle applies to individual talent. The people who materially alter the trajectory of an organization are rarely smoothly competent generalists. They're usually people with one or two conspicuous strengths—capabilities so distinctive that they change what the organization around them becomes capable of doing. But this competitive landscape for both founders and exceptional talent is being flattened at speed. The ability to simply do things—to write, to code, to design, to analyze, to build presentations—is being democratized. Abilities that once signaled genuine expertise increasingly look like basic competence. And "good enough" will no longer be good enough for anyone trying to stand out.

Commentator and mathematician Adam Spencer, speaking recently on the StoryWork podcast, offered a useful example of what a real spike actually looks like in practice. Most people assume his spike is obvious: he's the big-brained math guy, right? Except that's not quite it. Put him in a room full of mathematics PhDs and, in his own words, he becomes "as dumb as a box of hammers." What actually makes him unusual is the combination of capabilities he possesses. He's unusually mathematical, but also forensically verbal and properly funny. He can understand a difficult idea, find the interesting thing hiding inside it, explain it to people who know much less, and make them laugh while doing it. He's spent a career bouncing between comedy, radio, mathematics, science, law, technology, and public speaking without fitting neatly into any of those boxes. Through his speaking, writing, and Substack newsletter focused on AI, he's become one of Australia's leading voices on artificial intelligence.

What Spencer articulated on the podcast is that expertise itself is changing. For a long time, being an expert meant possessing knowledge or capabilities other people simply did not have. You knew how to code. You could speak another language. You understood the law. You could build financial models, fix software bugs, or analyze data. AI is increasingly putting a layer of those capabilities into everybody's hands. So if the answer itself becomes the cheap bit, knowing which question to ask becomes more valuable. If the machine can produce a plausible analysis, recognizing whether it's actually any good matters more. If everybody can build the presentation, the thinking underneath it matters more. And if AI can sound convincingly expert on almost anything, knowing where its expertise ceases to be credible matters more too. Spencer calls one version of this new spike "expertise midwifery"—not pretending to be the final authority on things, but making people curious enough about a difficult idea to want to understand it, then pointing them toward the real sources of expertise.

The real differentiator, Spencer suggested, may lie in the unique and odd combinations of strengths people bring. Maths alone is not his spike. Neither is comedy, public speaking, scientific curiosity, or his growing expertise in AI. His spike is the peculiar way those capabilities compound—the ability to be smarter than most funny people and funnier than most smart people. The same logic applies to companies. If similar foundational AI models are available to everybody, if everyone can code faster and analyze more data, then simply being "AI-powered" will rapidly become about as distinctive as being "internet-enabled." The interesting businesses will still need spikes, but those spikes may increasingly come from combinations competitors find difficult to assemble: extraordinary technology combined with brilliant real-world brand experiences, deep technical capability combined with a genuinely loved brand, data-driven decision-making combined with obsessively non-scalable customer service.

What bothered Spencer most about OpenAI's recent demonstration of GPT-6 Astra wasn't that the technology looked weak. It was that people had been handed astonishing new capabilities and mostly appeared to be using them to become more efficiently average. A man discussed creating a video game with the excitement of Michelangelo painting the Sistine Chapel. A woman became animated discussing an eBay listing with her computer. Another person asked AI to make a presentation "pop" while swinging a tennis racket around. Someone else got AI to do a chunk of his legal job for him. None of this was useless—removing drudgery and extending useful capabilities to more people is a genuine benefit. But the mismatch between the astonishing scale of capability on display and the modesty of the ambition being pursued was striking. As Google DeepMind founder Demis Hassabis argued at Cannes this year, while we may be overestimating AI's short-term impact, we are almost certainly underestimating how transformative it will be over the long term.

AI is raising the floor beneath all of us. For talent, that means capabilities that once differentiated you will increasingly become basic equipment. For founders, technologies and operating advantages that feel extraordinary today may be broadly accessible remarkably quickly. None of this diminishes the value of excellence. It makes distinctiveness more important because competence itself is becoming less scarce. At Sydney Contemporary, surrounded by thousands of competent, accomplished, and often beautiful things, the works you remember are the ones that somehow refuse to dissolve into the whole. The same may become true of almost everything else. AI is making it dramatically easier for all of us to put our names and our businesses up on the wall. The interesting question is what will make our talents and ideas catch the eye.