An Atlantic Argument Says AI Doesn't Have to Repeat Automation's Mistakes. Here's the Case, and the Holes In It

The Atlantic published an essay arguing that generative AI could avoid repeating the mistakes of the last 40 years of automation, if companies and policymakers deliberately choose to build what the author calls "pro-worker AI." The piece is framed as a call to action, not a report on something already happening. The argument and the evidence deserve separate consideration.

The core claim: over the past four decades, digital automation boosted productivity but concentrated the gains. Wages stagnated for large swaths of workers even as output per worker climbed. The Atlantic ties this to broader instability in liberal democracies, though it doesn't cite specific data in the excerpt to quantify that link.

That inequality trend itself is well-documented elsewhere. Multiple economists across the political spectrum, including at the Federal Reserve and in academic labor economics, have tracked a widening gap between productivity growth and median wage growth since the 1980s. Where they disagree is on causes. Some blame automation and trade. Others point to declining unionization, tax policy, or occupational licensing. The Atlantic's essay picks automation as the primary driver without engaging those competing explanations.

The Pro-Worker AI Pitch

The essay's central idea is that generative AI, unlike earlier automation, can augment workers rather than replace them. It cites computer scientist J.C.R. Licklider's 1960 prediction that computers would eventually enhance human cognition by delivering context-sensitive information on demand. The author argues large language models finally make that possible.

The practical example given is electricians. Instead of a general-purpose chatbot, the argument goes, a specialized AI trained on niche technical data, past repair cases, and site-specific sensor readings could help tradespeople diagnose equipment problems faster. That's a plausible use case. Specialized diagnostic AI already exists in some industrial and medical settings, and there's no technical reason it couldn't extend to electrical trades or other skilled labor.

The claim that this would "increase wages, employment, and productivity all at the same time" is the essay's most aggressive assertion, and it's presented as a prediction, not a proven outcome. No study or dataset is cited to back that specific triple-win claim. It's a hypothesis about what could happen if AI is deliberately designed and deployed a certain way.

What the Argument Leaves Out

Here's the concern a reasonable skeptic, whether pro-labor or pro-business, should raise: none of this happens automatically, and the essay doesn't fully grapple with why companies would choose the worker-augmenting path over the worker-replacing one.

Businesses adopt technology based on cost and return, not ideology. If a company can cut headcount using AI, many will, because that's the economically rational move under current market structures and shareholder pressure. The Atlantic essay acknowledges this transformation "won't happen on its own," which is an honest admission, but it stops short of laying out what specific policy, regulatory, or market mechanism would force firms to choose augmentation over replacement.

This is a legitimate structural question, not a partisan one. Free-market advocates would argue government mandates on how companies deploy AI risk slowing innovation and raising costs. Labor advocates would argue that without some intervention, whether through tax incentives, labor law, or collective bargaining, companies will default to the cheapest headcount-reduction path every time. Both concerns are reasonable and the essay doesn't resolve either.

Where This Actually Stands

As of today, there's no broad industry standard or regulation dictating that AI tools must be designed to augment rather than replace workers. Companies from manufacturing to customer service to law firms are making individual decisions about AI deployment based on their own cost calculations. Some tools genuinely do augment skilled workers, speeding up diagnostics or drafting. Others are explicitly marketed as ways to cut staffing.

The Atlantic's framing treats "pro-worker AI" as a coherent, buildable category that just needs societal will. That's an argument for what should happen, not a description of an existing trend or policy already in motion. No legislation, corporate pledge, or industry coalition cited in the piece currently mandates this approach.

The unresolved question is concrete: which companies, unions, or lawmakers are actually going to build the incentive structure that makes "pro-worker AI" the default rather than the exception? Until that mechanism exists, the choice between augmentation and replacement will keep being made quietly, deal by deal, inside individual companies, with no public accounting of which path they chose or why.