AI policies work better when employees help write them - Computerworld

TEXT START: Organizations that listen to employee concerns and codify worker protections are more likely to realize AI’s benefits without eroding trust.

The Dissection

This is transition management disguised as employee-centered governance. It correctly identifies fear, surveillance, botsitting, workslop, workload inflation, and weak training as barriers to adoption. But its real objective is to make AI deployment acceptable and operationally smooth. Employee input is treated as consultation to improve compliance—not as control over capital, deployment, staffing, or the distribution of gains.

The Core Fallacy

The article confuses procedural consent with economic security. Human review of employment decisions, training programs, and guardrails do not change AI’s cost and performance advantage. A human signature on a machine-generated redundancy is still redundancy. Under P1, AI continues to compress cognitive labor; under P2, policy cannot preserve human-only economic domains at scale; under P3, training does not guarantee economically necessary work. The article’s central promise—AI being done “with workers rather than to them”—is mainly packaging for worker cooperation during their own displacement.

Hidden Assumptions

• Productivity gains will be shared with workers rather than converted into fewer employees and higher output quotas.
• Training will create durable roles instead of producing better-qualified surplus labor.
• Human oversight will preserve meaningful worker discretion rather than merely provide legal cover.
• Corporate policies will restrain management when competitive pressure rewards faster labor substitution.
• Trust and communication can solve what is fundamentally a power and ownership conflict.
• Guardrails will prevent exploitation rather than simply make exploitation safer and more defensible.

The article even documents the contradiction: AI saves time, then employers increase workloads. That is not a policy glitch. It is the incentive structure functioning correctly.

Social Function

Transition management, ideological anesthetic, and partial truth. The proposed protections may reduce reckless deployment, discrimination, security failures, and workplace abuse. But the article converts a distributional conflict into a communications problem. It asks labor to help write rules for the process that is dismantling labor’s bargaining power.

The Verdict

Useful as an operating manual for deploying AI without triggering immediate resistance, unionization, or organizational breakdown. Useless as a defense of the post-WWII employment circuit. It may delay social death and improve adoption; it cannot prevent mechanical death. This is padded scaffolding around the severing of the mass employment–wage–consumption loop.