AI is stifling collective innovation: three tips for founders and investors
This opinion piece argues that heavy reliance on AI tools is making Danish startups more alike and weakening the ecosystem's ability to produce real breakthroughs. It is written by Jonatahn Løw, a former AI entrepreneur who writes and speaks about AI, and he stresses that he is not against the technology.
He starts from what founders tell him: AI tools are excellent at writing code overnight, automating B2B outreach, generating SEO and GEO content and assembling pitch decks, which saves time and raises the floor for how fast an idea can reach the market. But he cites research from MIT and Harvard showing that the more people lean on standard digital solutions, the more uniform and predictable their ideas, product features and go-to-market strategies become. The same research finds that AI raises the average quality of work, especially for people who are weak in a given discipline, so a technical founder can produce a decent market analysis and a commercial founder can generate working code.
The downside appears at ecosystem level. Language models are trained on historical data and return the statistically most likely, "safe" answer. When hundreds of founders and angels use the same tools to validate demand or plan growth, they end up with the same pricing models, LinkedIn growth hacks and pitches. Løw also warns of cognitive laziness and says AI is good at optimising the known, such as cutting server costs or tuning conversion rates, but major breakthroughs have come from the unexpected. His example is Slack, which began as a failed computer game whose team built an internal chat tool that was later sold to Salesforce for 27 billion dollars.
He offers three tips. First, keep AI out of the room in "phase 0", when the vision or a major strategic move is being shaped, using pen, paper and whiteboards, and use AI afterwards to stress-test numbers. Second, give prompts constraints, for example asking for scaling ideas that exclude the three most common playbooks, or a critique from a cynical 2008-era VC. Third, investors should deliberately reward unconventional go-to-market models and odd product architectures instead of discarding them as errors in standard screening templates.