Lecture 09 — Slop at work

Workslop, the office's own version of the problem, and why it hides in plain sight

19 April 2027

Everything so far has treated slop as a public, adversarial problem — someone out there trying to fool someone else. “Workslop” is quieter and, in a way, more corrosive: a 2025 Stanford/BetterUp study found that a large share of knowledge workers had received AI-generated material at work in the past month that looked complete but wasn’t — a report with no real analysis behind it, a summary that restated the brief instead of advancing it. Nobody was trying to deceive anybody. The generating cost of looking finished had simply dropped below the generating cost of actually finishing.

The forensic skills from earlier weeks still apply, but the stakes and the social context are different: you are not unmasking a stranger’s fraud, you are naming a colleague’s shortcut, inside a relationship you have to keep working within tomorrow. This session is about that harder conversation as much as it’s about the detection.

Outline#

• what the research actually measured, and what it didn’t
• why workslop is invisible to the sender and expensive for the receiver
• naming it without accusing a person: separating the artefact from the motive

Reading#

The Stanford Social Media Lab / BetterUp “workslop” research (2025) and the Harvard Business Review follow-up (2026) on why people produce it.

Further reading#

CNBC’s coverage of the workslop research is the fastest way into the study’s actual numbers — the $186-a-month figure and the reputational-cost findings this lecture cites both come from here.