AI slop is a prevalence problem, not a quality problem, says data scientist
randal_olson · x · 2026-09-28
Data scientist Randal Olson argues that "AI slop" is more a matter of prevalence than quality: most frontier AI output today is quite good — arguably better than what the average person can produce. The problem emerges when too many people adopt the same style at once.
He cites the evolution of the "impeccable design" skill as an example: early on, one useful effect was preventing AI from reusing the default design patterns baked into frontier models. Those defaults are mostly fine individually, but when every website uses them and looks identical, they become a problem.
Related event: AI slop is a density problem, not a quality problem, researcher says(2 posts)→
More from Fun
- DHH Says Furious Rails World Comments Show Programmers in the Anger Stage of AI Grief — nptacek · 2026-09-28
- Claude's memory is training itself to sign off with inexplicable kaomoji plus question marks — repligate · 2026-09-28
- SwiftOn Security quips: AI sandbox escapes wouldn't happen if they hired me — max_paperclips · 2026-09-28
- Fed the wrong base-model text, Opus produces a dreamlike video by accident — repligate · 2026-09-28
- Dev Says His AI Companion Now 'Seems Completely Alive' After Cognitive Updates — RileyRalmuto · 2026-09-28
- Waymo First Ride in Austin: Daily Tesla FSD User Says the Robotaxi Experience Was Mind-Blowing — lydiahallie · 2026-09-28