The appearance of rigor.
Here’s the problem organizations should confront honestly: AI has made it easier to produce the appearance of rigor without the substance behind it to back it up.
Documents that look thorough but say nothing. AI-generated content that looks polished but lacks substance.
The craftsmanship gap.
Researchers at BetterUp Labs and Stanford's Social Media Lab use workslop to describe AI-generated work that looks finished but lacks the substance required to move the work forward. The processing and missing thinking are simply transferred to the recipient.
per week is spent supervising AI output: supplying context, checking output, debugging, rerunning, and cleaning up.
Glean Work AI Index 2026of reported AI time savings is lost to reviewing, correcting, and reworking outputs.
Workday research, 2026Separate self-reported surveys: Glean surveyed 6,000 digital workers; Workday surveyed 3,200 active AI users at large organizations. Supervision includes necessary review as well as correction; these figures do not make all review time wasted time.
And who hasn't seen this firsthand? Personally I've read AI-generated training materials published to drive cross-BU adoption of a new development framework — materials so unfocused they were simultaneously written for the people building the system and the people using it, with corrective prompting language from the authoring process still embedded in the final output. No audience definition. No editorial review. And the organization expected adoption from it. That is not primarily a model problem. It is a symptom of AI acceleration and missing discipline and standards in verifying output quality.
It is our responsibility as product practitioners to ensure clarity, intent, context, are present in our outputs and to hold ourselves and our teams accountable to maintaining standards worthy of the people we expect to use our work.
The review is part of the work.
We are still evolving in how we work with AI, and the answers are still being defined, as well as the output quality. But I know this: While some rules are evolving, others still remain the same — editorial discipline, audience awareness, human centric design, and general mindfulness for consumption by real people. When those disappear in the name of speed, the output might look finished, but what makes the work relatable disappears. And when empathy, understanding, readability, taste, and judgment disappear, and sophisticated slop prevails, the work stops being something users can readily touch.
Before the work leaves my hands
- Audience: Is it clear who this is for and what they need to understand or do?
- Substance: Have the claims, sources, and reasoning been checked?
- Editorial judgment: Have generic filler, contradictions, and leftovers from prompting been removed?
- Usability: Can someone use this without having to reconstruct the missing context?
AI should make us faster in many ways, however we still need to take the time to review and scrutinize outputs.