AI, meet CBA - NW Labor Press

TEXT START: Tacoma News Tribune reporter Kristine Sherred couldn’t believe what she was seeing.

The Dissection

This is a labor-mobilization document disguised as explanatory journalism. It catalogs genuine failures—AI-generated editorial degradation, clinical scheduling errors, surveillance, opaque performance metrics—and converts them into a case for union bargaining, disclosure, human review, and limits on substitution.

Its strongest move is identifying AI’s blast radius: centralized systems can multiply one error across an institution. Its weaker move is treating implementation rules as if they determine the economic endpoint. The article describes how workers can contest deployment, but not how they retain productive necessity once the systems become cheaper and more capable.

The ending exposes the transition already underway. The publication rejects AI-written articles but accepts AI transcription, headline suggestions, metadata generation, and search summarization. Tasks are being peeled away in layers. The boundary called “human journalism” is already becoming a shrinking supervisory and credibility function.

The Core Fallacy

The article confuses control over AI’s introduction with control over AI’s structural consequences.

Union contracts can delay substitution, impose notice requirements, preserve severance, limit surveillance, and force human review. They cannot permanently create a human-only economic domain if AI achieves superior cost and performance. Under P1, employers that are restrained by one agreement can outsource, automate elsewhere, reduce the bargaining unit, close the operation, or redefine the work.

The article also treats current AI failures as evidence against eventual displacement. They are evidence of immature or badly deployed systems. A scheduling failure and a dull rewrite do not defeat the thesis; they create pressure for better systems, tighter verification, and cheaper automated correction. Human review may survive as a thin accountability layer while the mass of production disappears.

Under P2 and P3, bargaining is a lag defense. It can redistribute the carcass and slow the killing. It cannot restore the wage-to-consumption circuit once most labor is no longer economically necessary.

Hidden Assumptions

• Employers will remain inside enforceable bargaining structures rather than evade restrictions through vendors, outsourcing, reclassification, or closure.
• Union leverage will grow faster than the economic incentive to eliminate or compress the work.
• “Human-in-the-loop” will mean substantial human authority rather than ceremonial signoff over machine output.
• Readers will pay enough for human production to preserve the underlying employment base.
• Journalistic relationships, interviews, judgment, and voice will remain difficult enough to automate that they support mass employment rather than a small verification elite.
• AI errors will remain frequent and expensive instead of becoming manageable through better models, workflows, and layered checking.
• Employers will share productivity gains through wages and hours rather than use them to reduce headcount and bargaining power.
• Transparency, disclosure, and authenticity will command a durable premium against cheaper synthetic content.
• Contracts can preserve tasks without merely preserving a smaller number of workers who supervise systems replacing everyone else.
• Human solidarity can overcome competitive pressure when rival firms, vendors, or jurisdictions adopt unrestricted automation.

Social Function

Primary classification: transition management.

Secondary classifications: partial truth, labor mobilization, and ideological anesthetic.

The article is not pure copium. Its documented cases are real operational and labor problems, and its proposed tactics can win concrete concessions. But it implies that aggressive bargaining can preserve human productive participation at scale. That implication is the sedative.

“Keep news human” is a defensible quality and legitimacy demand. It is not a viable macroeconomic defense. A human byline can preserve trust while the platform, search engine, publisher, and distribution layer capture the revenue. The article itself admits that AI summaries can eliminate visits to the source, yet treats continued access and reader goodwill as sufficient strategy. That is moral purpose standing in for a business model.

The Verdict

This is a competent map of the first trench, not a solution to the war. Unions can force notice, extract transition rents, protect standards, and preserve narrow roles for accountable humans. They cannot repeal P1, defeat P2, or prevent P3 indefinitely.

The article sees the machinery entering the workplace but mistakes the emergency brake for a new engine. Its contracts are temporary moats around shrinking islands of human labor. When AI becomes reliably cheaper and good enough, “human review” will remain for liability, trust, and prestige—but the mass employment system that made those professions economically broad will already be dead.