What If 90% Is Good Enough? The AI Slop Apocalypse

A developer’s existential dread about the future of software craftsmanship—and whether anyone will care when it dies.

The Core Insight

Here’s the nightmare scenario: AI stops improving tomorrow. What do we have?

Models that can almost code a web browser. Almost code a compiler. Can demo pretty well if you let them take shortcuts. Can make you 90% of a thing.

“90% is a lot. Will you care about the last 10%?”

This is the question haunting software developers in 2026. Not whether AI will replace them—but whether the “good enough” tide will erode the very concept of software quality until nobody remembers what they’re missing.

Why This Matters

The Craftsmanship Crisis

The author isn’t worried about AI agents writing apps they’ll never experience. They’re terrified of the humans—the “AI herders” who won’t care enough to understand what they ship. And the users who’ll consume the slop and be fine with it.

It’s an uncomfortable parallel: they’ve made peace with IKEA furniture. But software temufication hurts more than commoditization. We’re not heading toward competent mass-market—we’re heading toward dropshipping hell.

The Median Magnet

AI tools nudge toward sameness. Try making something unique like Paper by FiftyThree—you’ll end up with “normal and uninspired” every time. Claude and friends can help learn new technologies and craft good software, but they constantly pull toward that same Next-React-Tailwind median.

Going off the beaten path? These things just don’t handle it well.

Slop Was Always Here

Let’s be honest: slop predates AI. Human decisions led to uncomfortable chairs, SEO-polluted search results, and ticket booking interfaces that induce tears. Incentives have never aligned to produce good software. “Move fast and break things” was a human philosophy.

AI just accelerates the pattern. Agent herders can now produce mediocrity at unprecedented velocity.

Key Takeaways

• The 90% trap: AI can deliver most of a solution—enough to ship, not enough to delight
• Learned helplessness: What if people genuinely don’t care about privacy, quality, or craft?
• The imagination gap: AI can’t be imaginative; developers can’t afford to be; will users fill the void?
• The torch problem: Who carries forward the knowledge of what good software looks like?
• Speed vs. craft: The future may belong to whoever churns fastest, not whoever builds best

The HyperCard Dream

There’s a hopeful counter-narrative: maybe AI tools bridge the gap between users and developers. ChatGPT as the new HyperCard. Everyone turning ideas into reality with a few sentences.

Evidence exists. Carol in Accounting builds insane Excel spreadsheets. Kids on TikTok automate phones with Shortcuts and hack up Notion notebooks. Non-programmers coding without knowing it.

But what if they’re the aberration? What if tech learned helplessness is terminal? What if most people really do just want “a glorified little TV in their pocket”?

Looking Ahead

The essay ends on a note of genuine dread:

“I’m terrified that our craft will die, and nobody will even care to mourn it.”

This isn’t a technical concern. It’s existential. The fear isn’t that AI will make software worse—it’s that software quality will decline and nobody will notice. That “good enough” really will be good enough for the market, and craftsmanship becomes an affectation rather than a professional standard.

Maybe it’s unwarranted doomerism. Maybe the cream will rise. Maybe quality will find its market niche.

But the question lingers: in a world where 90% of a solution can be generated in seconds, who will pay the premium for the last 10%?

Based on analysis of “(AI) Slop Terrifies Me” from ezhik.jp

Tags: #AISlop #SoftwareCraft #QualityVsQuantity #DeveloperCulture #TechPhilosophy