Want Employees to Embrace AI? Stop Selling It as a Productivity Tool | Stanford HAI

TEXT START: How managers introduce and support AI determines whether workers engage with or resist the technology.

The Dissection

This is a managerial adoption manual disguised as worker-centered research. It does not ask whether AI preserves the worker’s economic necessity. It asks how to make threatened workers use the instrument that is eroding it.

The law-firm comparison demonstrates a real mechanism: credible training, protected experimentation time, and expanded responsibilities increase adoption. But the expanded role is a staging area, not a stable destination. “Job enrichment” is the language used to make displacement acceptable while the organization learns which tasks can be removed.

The article also exposes the underlying machine. It admits that weakly bundled jobs—paralegal work and transcription—are easier to disassemble. It admits that promised time savings are usually captured by raising output expectations. That is not an implementation glitch; it is competitive logic.

The Core Fallacy

It conflates successful AI adoption with successful human adaptation. A worker can use AI more often, perform more sophisticated tasks, and gain temporary status while the firm maps the remaining tasks for automation. Engagement measures compliance and experimentation, not durable bargaining power or productive necessity.

Its second error is treating higher-level work as a refuge. Once AI makes routine research, drafting, and analysis cheap, the “interesting” layer becomes the next target. P1 drives cognitive cost and performance superiority; P2 prevents firms from maintaining large protected human-only domains; P3 follows as economically necessary labor contracts. The article’s proposed enrichment is therefore a lag defense, not a reversal.

Hidden Assumptions

• Firms will share AI-created surplus with workers instead of converting time savings into higher quotas, fewer staff, or lower labor costs.
• New responsibilities will expand faster than AI can absorb them.
• Human judgment, relationship work, and meaningful tasks remain scarce rather than becoming partially automatable and standardized.
• Metrics can be redesigned to recognize value without triggering the same competitive pressure that destroys the role.
• Training and experimentation create durable careers rather than making workers more efficient during a transition to a smaller workforce.
• Individual managers possess enough discretion to override the firm-wide imperative to reduce labor input.
• Employee resistance is mainly a framing problem rather than a rational response to declining job security and bargaining power.

Social Function

Transition management and ideological anesthetic, with a substantial partial-truth component. The evidence about framing, scaffolding, and credible support is real. Its ideological function is to relocate the conflict from ownership and employment to managerial communication: if workers resist, the rollout was mishandled; if they comply, the system appears humane. It teaches institutions how to secure voluntary participation in their own labor substitution.

The Verdict

The article is a polished hospice protocol for cognitive labor. It can improve AI adoption, preserve dignity temporarily, and create servitor niches around verification, relationships, and system integration. It cannot preserve mass productive participation. Under the Discontinuity Thesis, “AI enrichment” is the soft wrapper around the hard event: the wage-and-consumption circuit is being severed, and better framing only makes the severing more orderly.