Everyone's using AI to brainstorm now. Need a startup idea? Ask ChatGPT. Stuck on a name for your project? Let Claude handle it. Can't figure out your next blog post topic? There's a prompt for that. And look, AI can give you ideas. That part is true. It's fast, it's fluent, and it'll spit out ten options before you've finished your coffee. But here's the thing nobody wants to hear: those ideas aren't really yours. And worse, they're probably the same ideas it gave a thousand other people this week.

Ideas are cheap, authenticity isn't

The dirty secret of AI-generated ideas is that they converge. Research from Wharton published in 2025 found that AI users tend to generate strikingly similar ideas, which can actually hinder innovation at scale. When everyone draws from the same well, the water starts tasting the same. A massive study from the University of Montreal, comparing over 100,000 humans against leading AI models like GPT-4, confirmed something important: AI can now beat the average human on creativity tests. But the top 10% of creative humans still leave every AI system far behind, especially on richer work like storytelling and poetry. The takeaway isn't that AI is uncreative. It's that AI creativity has a ceiling, and that ceiling is mediocrity dressed up in good grammar. This tracks with what I've seen firsthand. The names AI generates? Absolute garbage. Always some variation of the same overused patterns. The ideas it suggests? They sound smart for about three seconds until you realize they're recycled from the same pool of things millions of other people have already prompted and received.

AI's training data is its prison

Here's why this happens. Large language models generate text by predicting the most likely next token based on patterns in their training data. That's it. They're not imagining new possibilities. They're remixing what already exists. This means AI literally cannot give you an idea it hasn't seen before. It can combine existing concepts in novel arrangements, sure, but it can't reach beyond the edges of what it knows. And just like how you and I don't know what we don't know, AI has the same blind spot, except it doesn't even realize it has one. A study published in PNAS Nexus tested the creativity of multiple AI models and found that while individual outputs could seem creative, the responses across models were "surprisingly homogeneous." The researchers warned that widespread reliance on AI for creative tasks could lead to a measurable loss of unique ideas across society. That's not a theoretical risk. That's already happening.

Zero-human companies are still a fantasy

Some people take the AI hype further and argue we're heading toward companies run entirely by AI, no humans needed. KPMG and the University of Amsterdam launched an experiment in late 2025 to test exactly this, building a "zero-person company" powered entirely by AI agents. The goal was to understand which processes AI could take over and where human expertise remains essential. The results so far? Humans are still essential. The experiment revealed that while AI agents can handle structured, well-defined tasks, they struggle with the ambiguous, contextual decision-making that running a real business demands. Meanwhile, a study from the National Bureau of Economic Research surveying 6,000 executives found that nearly 90% of firms reported AI has had no meaningful impact on employment or productivity over the past three years. Companies are definitely cutting jobs and attributing it to AI, don't get me wrong. Block laid off 4,000 employees in early 2026 citing AI automation. But cutting headcount and actually replacing human judgment are two very different things.

The real problem with idea generators

Let me be blunt about "AI idea generators" specifically. They're bullshit. Not because the technology doesn't work, it technically does what it promises. You type in a prompt, you get ideas back. The problem is that the entire premise is wrong. Good ideas don't come from generators. They come from lived experience, from noticing patterns nobody else has noticed, from the friction of doing real work in the real world. The Harvard Gazette reported on research suggesting that excessive reliance on AI-driven solutions may contribute to "cognitive atrophy," a weakening of critical thinking abilities over time. When you outsource your thinking to a machine, you don't just get worse ideas. You get worse at having ideas. And that's the trap. The more you lean on AI for creativity, the less creative you become, which makes you lean on AI even more.

Pick up the camera and shoot something

I'm not saying don't use AI. I use it. Most people do now, and there are genuine productivity gains to be had for structured tasks, analysis, summarization, and first drafts of routine work. But when it comes to the stuff that actually matters, the ideas that define your business, the creative vision behind your content, the name that makes people remember you, do the work yourself. Come up with something original. I guarantee you it won't come from ChatGPT or Claude, because originality by definition can't come from a system trained on everything that already exists. The most creative humans still outperform AI by a wide margin. That's not a consolation prize. That's the whole game. Authenticity, taste, and original thought are becoming the scarcest resources in a world flooded with AI-generated content. The people and businesses that lean into what makes them human will be the ones that stand out. So pick up the camera and shoot something. Write the thing nobody asked for. Build the product that doesn't fit neatly into a prompt template. The best ideas have always come from people who were willing to be wrong in interesting ways, and that's something no model can do.

References