What is the most dangerous AI behavior at work?

The real AI risk is not that people will stop working. It is that they will produce more “work-shaped” material that shifts the burden to others. BetterUp Labs and Stanford Social Media Lab call this “workslop”: AI-generated content that looks finished, but lacks substance.

It may be a neat summary, a long deck, a clean email, or a tidy analysis. But if colleagues must decode, verify, correct, reframe or redo it, the productivity gain is fake. Worse: the person using AI saves time by spending someone else’s. That is not efficiency. It is cognitive dumping. In the AI age, the critical soft skill is not just prompting. It is knowing when your output is actually useful to the team.

1. Ban “polished but empty” work. Do not reward volume, speed or formatting. Reward clarity, usefulness and decision-ready thinking.

2. Ask one brutal question before sharing AI-assisted work. “Will this save my team time — or quietly create work for them?”

3. Make AI use a team norm, not a personal shortcut. Define what “good enough to send” means: checked facts, clear context, explicit assumptions, and a real recommendation.