AI Is About to Make Good Ideas More Valuable
For most of business history, ideas have been cheap and execution has been expensive.
You could have ten ideas for an ad campaign, but somebody had to write the ads, design the creative, build the landing pages and pay to test them. You could imagine a new product, but somebody had to research it, build it, name it, package it and take it to market. You could think of five ways to position the company, and testing all five might take months.
So businesses learned to suppress ideas. Pick one, build it, give it time, stop getting distracted.
That equation is changing faster than most owners have noticed, and it points somewhere counterintuitive: the value of having an unusually good idea is about to go up.
Two Competitors, Same Tools
The first uses AI the way most businesses currently do. Write this email. Make this ad better. Give me ten social posts. Rewrite this landing page. Summarise this meeting.
The business gets more productive. Marketing improves. Work that took five hours takes twenty minutes. That is genuinely valuable, and the competitor has the same tools.
The second owner uses them differently.
She has five separate theories about why customers are not buying. Rather than debating them for three weeks, she builds five landing pages and tests them. She notices an underserved segment and has AI research it overnight. She thinks of a strange new offer and has a prototype by tomorrow. She develops ten genuinely different ad concepts instead of ten variations of the same one. She finds something that works and generates fifty ways to extend it.
The first owner got faster. The second got creative leverage, and the distinction is about to matter a great deal.
Tomorrow’s forecast, in your inbox.
One email. Five minutes. Written for owners, not engineers.
AI May Not Make You More Creative
There is evidence that AI improves creative output. In a study published in Science Advances, people given access to generative AI produced stories judged more creative, better written and more enjoyable, with the largest improvements among people who scored lower in creativity to begin with.
There was a catch. The AI-assisted stories became more similar to one another.
Research in Nature Human Behaviour found the same shape in group brainstorming: AI raised the average quality of ideas while reducing the diversity of ideas a group produced.
Which is an awkward finding for anyone competing on marketing. If every company uses AI to generate its ideas, AI makes everyone’s marketing better and everyone’s marketing more alike at the same time.
You have already seen it. The polished LinkedIn post, the well-structured email, the clean website, the competent ad. Nothing wrong with any of it, and nothing memorable either.
Competence stops being an advantage the moment competence becomes nearly free.
Originating Versus Selecting
Ask for ten ideas to market your business and the machine originates while you select from what it produced.
Walk in with something else — I think our customers do not want X, I think they want Y, and every competitor is selling X, help me work out whether I am right — and the roles reverse. You originate. The machine amplifies.
That sounds like a philosophical distinction. It determines how much money AI makes you.
Take paid advertising. An owner normally has one theory about why someone would buy. She creates an ad, sends traffic to a landing page, waits to see what happens.
With AI she can start with five genuinely different hypotheses, build five ad concepts and five matching landing pages, and put all of it in front of the market. One gets almost no response. Three perform about the same. The fifth produces twice the qualified interest of everything else.
What she has produced is information about what her market actually wants, on top of the advertising itself, and the next experiment gets built around that signal.
What would once have taken an agency weeks and thousands of dollars is now an afternoon.
The Cost of Being Wrong
Companies have talked about testing for decades. In practice testing has been expensive — a proper test needed a strategist, a copywriter, a designer, a developer, an analyst and a media buyer. Most small businesses did very little of it. They guessed.
AI lowers the cost of being wrong, which means a business can now explore ideas it would previously have discarded for being too expensive to investigate. What if we changed the market. What if we sold the opposite benefit. What if the thing we treat as a side feature is the reason people buy.
Most of those fail, which is the point.
If testing an idea costs $20,000 you have to be careful about which ideas you test. At $200 you can afford to be curious, and curiosity occasionally finds something enormous.
What Creativity Is Worth Now
The popular assumption is that AI reduces the value of human creativity, since machines generate ideas too. I think that misses what happens when the cost of execution collapses.
AI reduces the value of average creative production. Another competent email, another competent ad, another competent website — all worth less than they were.
The unusual insight that makes one of those outperform everything else is worth more, because it can now be exploited immediately. A good idea no longer sits in a notebook waiting for budget, staff or time. It becomes an experiment today, and if the market responds it becomes a hundred experiments tomorrow.
Your AI Advantage Is Not the AI
Within a few years every serious business will have access to extraordinarily capable models. Your competitor will have them too, which makes “we use AI” worth roughly nothing as a position.
The advantage comes from what you put into it. What you noticed. What assumption you questioned. What connection you made that nobody else did. What the market told you that everyone else overlooked.
The machine can then research it, challenge it, build it, multiply it and help you test it.
The Forecast
As execution approaches free, the scarce input becomes the idea worth executing.
Over the next few years the businesses that pull ahead will be the ones running the most experiments rather than the ones producing the most output. Same tools, same budgets, and a widening difference in how many genuinely different bets get placed.
Expect the convergence to get worse first. As more companies generate their marketing the same way, the sameness deepens, and the businesses willing to test something odd will stand out for reasons their competitors cannot easily copy.
Medium-high rather than high, because the research on creative homogenisation is early and the studies measure stories and brainstorming rather than businesses. What I am confident about is the arithmetic. When the cost of testing an idea falls by two orders of magnitude, the constraint moves from execution to having something worth testing.
Which gives a different operating principle for the next few years. Rather than asking AI for more ideas, use it to make finding out whether your own ideas are right dramatically cheaper.
The biggest financial opportunity here may turn out to be affording fifty cheap failures on the way to the one idea that pays for everything, rather than doing the same work faster.
Sources: Doshi AR, Hauser OP. Generative AI enhances individual creativity but reduces the collective diversity of novel content. Science Advances, 2024 · Research on generative AI and idea diversity in group brainstorming, Nature Human Behaviour, 2025