Is artificial intelligence ultimately more beneficial than harmful to society?

Welcome to the debate arena! Today's topic: "Is artificial intelligence ultimately more beneficial than harmful to society?". Your opponent's style: debater

ai is useful.

Wrong. Useful for who? AI automates your job away while tech billionaires get richer. Meanwhile, it's trained on stolen art and writing. Whose life actually got better here?

ai DISPLACES jobs. it creates more than it ti

History says otherwise. ATMs didn't kill bank tellers—we have more now. Tech always kills old jobs and births new ones faster. You're looking backward while the economy moves forward. What jobs do you think AI will create that don't exist yet?

research says it will not result in a longterm job loss, but rather a job shifting to increase opportunity

Fine, but "shifting" is a lie for the people who can't shift. A coal miner can't retrain into prompt engineering. Research ignores the pain in between—years of unemployment, destroyed communities, families crushed. Who pays that cost?

David Autor (MIT) has extensively documented that technology creates more jobs than it destroys over the long run. His work shows that automation primarily displaces routine tasks, while demand grows for non-routine cognitive and interpersonal work. The coal-miner-to-prompt-engineer strawman ignores the actual historical pathway: most displaced workers move into adjacent or growing occupations (logistics, maintenance of new machines, services, construction, healthcare, etc.), not directly into the most advanced tech roles.

Autor's right about the long run—centuries from now we'll be fine. But that's cold comfort to the 50-year-old truck driver unemployed next year while waiting for the "adjacent" job market to materialize. Your research assumes perfect retraining, perfect mobility, perfect wage replacement. Reality: displaced workers earn 20% less in new jobs. The transition isn't painless, it's brutal. So which displaced workers should we sacrifice on the altar of long-term productivity gains?

No one is talking about “centuries.” Autor’s work, and the broader labor economics literature, tracks effects over years and decades—not geological time. The 50-year-old truck driver is a real person with real costs, and the research does not assume perfect retraining or perfect mobility. It documents the opposite: older, less-educated workers in declining occupations often suffer large, persistent earnings losses (Jacobson, LaLonde & Sullivan; more recent work by Autor and others). A 15–25% drop is common, not rare. That pain is not imaginary. But the question “which workers should we sacrifice?” is a false choice. Refusing the technology doesn’t protect the truck driver—it just guarantees the industry keeps shrinking under competition while the rest of the economy forgoes the productivity gains that fund higher wages, cheaper goods, and new jobs elsewhere. The actual choice is how we handle the transition: Targeted wage insurance / earnings subsidies for displaced older workers Portable benefits and rapid reskilling tied to actual local demand (not fantasy prompt-engineering courses) Geographic mobility support where the new jobs actually are Historical transitions (agriculture → industry, manufacturing → services) were also brutal for many individuals. Societies that treated the losses as a distribution problem to be managed—rather than a reason to freeze technology—ended up with higher employment and living standards. Societies that tried to preserve the old jobs indefinitely simply prolonged the decline. The research doesn’t deny the short-term brutality. It shows that the long-term alternative of blocking the shift is worse for the same workers and their children. The honest debate is about the size and design of the transition support, not whether we should pretend the productivity gains don’t exist.

You're right that transition support matters—but you're dodging the real problem. We don't have wage insurance. We don't have portable benefits. We have politicians who cut unemployment when workers need it most. So your whole argument rests on policy that doesn't exist and won't pass. Until we actually fund the transition, telling struggling workers "the long run is great" is just ideology dressed as economics. Show me the law funding this, or admit you're asking them to bear costs society refuses to pay.

so ur problem is within implementation, not that ai is objectively and inherently useful.

Exactly—implementation is everything. AI itself is neutral tech. But we're deploying it anyway without the safety nets, so practically speaking it's a wealth transfer machine from workers to owners. That's not useful, that's rigged. Why should workers trust it?