Ryo Lu argues that software makers should treat their work as an artistic medium, with intention and lived experience shaping product decisions instead of letting metrics or competitors do the choosing.
In an essay posted on X, Lu describes a familiar loop: teams copy language, aesthetics and features that already seem to work, then call the result learning from the market or following best practices. The products may look polished, he writes, but they also become interchangeable.
How AI accelerates convergence
Lu argues that measurable attention can gradually take over the creative process. Feedback stops informing the work and starts deciding what gets made, shifting the central question from what a maker wants to express toward what the market is likely to reward.
AI makes that cycle easier to repeat, in Lu's view, because it can draw from more references than any person could encounter. But access to many perspectives is not the same as having one. Without a person supplying an intention, he says, the result tends to be plausible and easy to approve. That output can then be shipped, measured and copied into the next round.
The concern extends to the people using the tools. Lu argues that makers may start describing themselves in categories that platforms can distribute and markets can recognize, pushing aside the contradictions that might have produced more distinctive work.
What an artistic approach would change
Lu's alternative is to build more like an artist: develop a perspective through making, noticing, choosing, failing and returning to the work. Personal experience matters because it gives a maker something specific to say.
Software, he argues, can express beliefs about how life should feel, what deserves attention, what users should be free to change and whether a task should feel like paperwork or play. That means a product decision should not be borrowed automatically from a competitor's roadmap.
Lu is not rejecting influence or faster tools. Artists learn from others too. His distinction is whether those influences become part of a point of view or substitute for developing one. If AI makes it possible to create more quickly, he concludes, deciding what deserves to exist requires more judgment, not less.
