New research says AI will raise wages for lower skilled workers more than anyone else

[Sharon79]

A new study from Stanford economist Lukas Althoff concludes that AI is likely to reshape jobs rather than eliminate entire occupations, and will ultimately raise wages across the workforce, with the biggest gains going to lower skilled workers specifically. His modeling estimates these workers could earn 15 to 45 percent more over their lifetime than they would without AI's support, with the gains potentially even larger depending on how quickly the technology continues to advance

The underlying mechanism is task simplification rather than task elimination. By reducing the specific skills required to perform many job tasks, AI lets lower skilled workers take on higher paying roles they previously couldn't access, which Althoff's modeling suggests would actually narrow the wage gap between the highest and lowest earners rather than widen it as many have feared

Not every path through education fares equally well in this framework though. For workers who do pursue a college degree, majors intensive in manual and technical skill, mechanics, engineering, culinary arts, cosmetology, architecture, transportation and physics, are projected to do better in the job market than degrees heavy in verbal and social skills, including political science, sociology, philosophy, law, religion and women's studies

Althoff notes that lower skilled workers have actually been slower to adopt AI so far, but not slowly enough to undo the technology's overall equalizing effect in his model. He describes the paper's contribution as translating now standard measures of AI exposure into meaningful wage and employment impacts for each individual in the economy, accounting for a worker's particular strengths and weaknesses, their capacity to learn new skills, and broader economy wide shifts in demand for goods and services, producing what he calls a more disciplined set of predictions about the AI economy than the field has typically offered

Skilled Workers Ai Worldphys.org

[Grover26]

The task simplification versus task elimination distinction is the whole ballgame here, and it's a more optimistic framing than most doom laden coverage of AI and jobs tends to offer

[Tracey49]

Verbal and social skill heavy majors doing worse feels like it cuts against the conventional wisdom that those are the safe humanities careers AI can't touch, worth sitting with that reversal

[DataProphet]

Lower skilled workers adopting AI more slowly but the equalizing effect still holding up in the model is an interesting wrinkle, implies the benefit doesn't require every worker to actively use the tools themselves

[Dave]

This is a much more grounded and quantified claim than the usual vague reassurance that AI creates as many jobs as it destroys, actual lifetime earnings estimates give it real teeth

[Ranger]

Would love to see this tested against real observed wage data over the next few years rather than staying a modeling exercise, predictions like this have a mixed track record historically

[DarkDan33]

Manual and technical majors outperforming verbal and social ones is going to be an uncomfortable read for a lot of current college students and their advisors

[Grim Socket]

The mechanism behind this makes some sense.

If AI tools lower the skill barrier for certain tasks, lower skilled workers can suddenly perform closer to higher skilled ones.

That compresses productivity gaps, which can translate into wage pressure upward.

Think customer support agents using AI to handle complex queries faster.

The output improves without years of training.

That is where the wage effect could come from.

[Megan34]

There is a historical parallel with spreadsheets.

Before them, financial modeling required specialized training.

Afterward, a much broader group could do meaningful analysis.

That did not eliminate experts, but it raised the baseline capability.

AI feels like a more general version of that shift.

So the wage uplift argument is plausible.

But it will depend on how widely tools are adopted.

[DarkMatter55]

The key question is bargaining power.

Even if productivity increases for lower skilled workers, wages only rise if employers compete for that improved output.

If the gains are captured entirely by firms, workers might not see much benefit.

So the institutional context matters a lot.

Not just the technology itself.

[Chris_50]

There is also a cultural factor.

Companies that view AI as augmentation may share gains with workers.

Those that view it purely as cost reduction may not.

Same technology, different outcomes.

That variability makes prediction tricky.

[Tel92]

Retail and logistics might be interesting test cases.

AI-assisted inventory management or scheduling could make frontline workers more effective.

If that translates into measurable productivity gains, wages could follow.

But those sectors are also highly cost-sensitive.

So it is not guaranteed.

[One-One-Five]

There is also a risk of role expansion without compensation.

If AI makes someone more productive, companies might simply assign more tasks rather than increase pay.

That has happened in past productivity waves.

So the outcome is not guaranteed

Depends on labor market dynamics.

[MJF_Fan]

There is a subtle distinction between tasks and jobs.

AI tends to automate tasks rather than eliminate entire roles.

If lower skilled roles consist of many automatable tasks, AI can reshape them into higher value positions.

That could support the wage increase argument.

But it will vary by sector.

[Taker29]

Another angle is training costs.

If AI reduces the need for long onboarding or formal education, more people can enter certain roles.

That increases labor supply, which could counteract wage growth.

So there are opposing forces at play.

Not a simple one-direction effect.

[Phil]

Some early examples are already visible.

Call center workers using AI assistants can handle more calls with better resolution rates.

Junior developers can ship code faster with AI copilots.

Those roles become more valuable at the margin.

The question is whether pay adjusts accordingly or just expectations increase.

[NoMercyElliot54]

The study is interesting, but real-world validation will take time.

Wage data lags behind technological adoption.

It might take several years before clear patterns emerge.

Until then, it remains a well-argued hypothesis.

Worth tracking though

[Rooney93]

The distribution of gains will likely be uneven.

Some lower skilled roles will benefit more than others.

Jobs that involve interaction, judgment, or adaptability may see bigger boosts.

Routine roles might not gain as much.

So averages can hide important differences.