We are currently drowning in a flood of frictionless content.
Feed a modern language model four sentences of prompt, and it will emit twelve verses of rhyming couplets that scan decently. Prompt an audio generator, and it will render a three-minute lo-fi beat with fake vinyl hiss and a smooth jazz Rhodes piano progression in twenty seconds flat.
Technically, none of it is broken. There are no dropped frames, no dissonant chords, no syntax errors.
And yet, you feel absolutely nothing while listening to it.
Why? Because the output was born without resistance. It is the average of five billion internet scrapings, smoothed out to remove any friction, idiosyncrasy, or risk. It lives in the uncanny valley of creative labor: polished, hyper-competent, and hollow.
As someone who spends his days writing TypeScript systems and his nights producing music, here is what I think about the machine in our studios.
1. Friction Is Where Taste Actually Lives
Every creative choice you make is defined by what you refuse to do.
When I’m tracking a song, my taste isn’t just picking a snare sample. My taste is the fact that I tried twenty different snares, hated all of them, got frustrated, turned down the microphone gain, and slapped my palm against a wooden desktop because it made a stupid dull thud that felt right for that specific bar.
A generative model cannot experience frustration. It doesn't have an aching neck from leaning into studio monitors. It doesn't know what it feels like to delete three days of code because the architecture felt bloated and dishonest.
When you remove all the friction from creation, you also remove the crucible where human taste is forged. If an algorithm generates a complete solution on the first try, you haven't made a choice—you’ve merely approved a prediction.
2. The Chisel, Not the Autopilot
I don't hate modern machine tools. I use them constantly to inspect ASTs, write boilerplate interface types, and refactor repetitive CSS tokens. They are phenomenal chisels.
The disaster occurs when people treat the chisel like a sculptor.
Good use: "Scan this 800-line Web Audio node graph for memory leaks in the cleanup cycle."
Bad use: "Write an album opener that feels like being stranded in a rainy Nordic city."
The first is mechanical offloading; it saves your working memory so you can stay in the zone. The second is an abdication of authorship. The second you ask a statistical predictor to tell you what your own grief or rage feels like, you have surrendered the only reason to make art in the first place.
3. The Reassertion of the Imperfect
The more synthetic content inundates our feeds, the more valuable the human flaw becomes.
In software, it’s why developers are falling back in love with handcrafted, idiosyncratic personal websites with weird custom physics and bold, non-standard navigation—subverting the homogenous Figma-component aesthetic that turned every web app into a corporate SaaS clone.
In music, it’s why a tape-distorted demo recorded on a cracked phone microphone cuts through the noise far deeper than a pristine, AI-generated orchestral master.
The future doesn’t belong to fully automated factories generating synthetic wallpaper. It belongs to the stubborn humans who know how to use the machine as a heavy hammer, while leaving their own distinct dents on the anvil.
