The Imagination Gap
Artificial intelligence is changing what developers imagine building, and the effect runs deeper than speed alone.
The cost of a software idea has always been paid in time. Building a working application from scratch takes weeks of skilled work at minimum, often months. Most ideas die in the gap between thinking of something and actually building it, because the cost is simply too high. Yet artificial intelligence is cutting that cost with surprising speed. When effort drops sharply enough, imagination changes.
I have spent fifteen years building web applications in JavaScript, long enough to have a reliable sense of what software costs in time. A complete photo editor, built from scratch in an environment I had never worked in, would normally take months. Using Claude, it took a few hours. I had no experience with native MacOS development. An AI assistant produced the basic structure and framework. The white-balance controls and bulk file-export were working before I had written a single line of code. The limit, I realised, had become my imagination.
This effect is the hidden tax on software creativity. Developers routinely rule out ideas based on how long they will take. Any project too large for available hours gets set aside before it begins. The filter is so automatic it rarely feels like a decision. AI does not remove it entirely, but it lowers the bar substantially.
The results are practical and concrete. Take turning a large dataset into a visual graphic. I wanted to produce historic timelines from large CSV files. Writing code to do this is better than drawing by hand, but only if the time spent writing the code is less than the time spent drawing. For a solo developer working evenings, those costs had been roughly equal, wiping out the advantage. With AI handling the interface and basic framework code, the coding approach becomes clearly cheaper. A whole category of solution becomes worth attempting.
The same logic applies to data tools, internal software, and experimental products that were never quite worth building before.
For companies, the implications go deeper than raw output. Engineering talent has long been one of the main advantages in the technology industry. If building becomes cheaper, the scarce resource shifts toward ideas. The ability to define what should be built, for whom, and why becomes the asset that matters. Product judgment, design sense, and specialist knowledge gain value as the cost of coding falls.
The shift carries real risks. When building becomes cheap, speed can get ahead of clear thinking, and software built quickly and without friction is still software that can fail, alienate users, or solve the wrong problem elegantly. The hard problems have moved. When I say that working with AI lets me “imagine a delightful experience” instead of a minimum viable product, I am describing access to a harder problem.
The imagination gap is closing. Whether wisdom closes with it remains to be seen.