It’s time to clean house — AI should replace some humans

### AI Could Boost Economic Growth While Disrupting Knowledge Jobs

Anthropic’s latest economic scenarios highlight a striking possibility for the future of work: artificial intelligence could dramatically increase economic productivity while simultaneously causing significant job losses.

The figure drawing particular attention is an estimated 11.9% unemployment rate by 2030 under the company’s most extreme scenario, where AI advances rapidly and becomes deeply embedded across the economy.

The impact could be especially severe for knowledge workers, including highly educated professionals who have traditionally viewed education, experience and specialized expertise as protection against technological disruption.

Under Anthropic’s extreme scenario, unemployment among knowledge workers could reach 17.9%, while employment across knowledge-based occupations could fall by more than 20%.

At the same time, the overall economy could become substantially larger than it would have been without AI, creating a paradox at the center of the technology revolution.

An economy can become significantly more productive while requiring fewer people to perform the work that previously generated that productivity.

Unlike earlier waves of automation, which largely focused on physical labor and repetitive manufacturing tasks, modern AI is increasingly capable of performing professional and cognitive work.

Software development, customer service, accounting, financial analysis, legal support, administration, research, marketing and content creation all involve activities that advanced AI systems can increasingly perform.

These systems can process information, identify patterns, generate written material, prepare documents and produce recommendations at speeds that are difficult for individual workers to match.

Humans may continue making final decisions in many professions, but the amount of human labor required to reach those decisions could decline substantially.

That prospect is particularly unsettling for workers who followed the traditional path of earning degrees, gaining professional qualifications and developing specialized expertise.

For decades, the underlying assumption was that advanced education and specialized cognitive skills would make workers more valuable as economies became increasingly sophisticated.

AI is now challenging that assumption by making some of the skills that once commanded a premium in the labor market increasingly accessible to machines.

However, the economic debate around AI often focuses heavily on the cost of replacing workers while paying less attention to the cost of human inefficiency.

When AI replaces a group of employees, the immediate questions tend to involve lost wages, unemployment and the cost of retraining those workers.

Those concerns are important, but another question deserves equal consideration: how much economic value was the work actually producing?

Employment figures generally do not distinguish between highly productive work and positions that exist because organizations rely on outdated, complicated or inefficient processes.

A worker processing paperwork that could be automated is still counted as employed, just as an employee performing redundant administrative checks remains part of employment statistics.

This issue becomes particularly significant in government, where inefficient systems can continue operating for years without the same competitive pressures faced by private businesses.

Delays, duplicated procedures, unnecessary layers of approval and administrative mistakes can impose substantial economic costs on citizens and businesses, even though those costs rarely appear in traditional employment calculations.

AI could potentially help identify these inefficiencies by analyzing large volumes of government decisions, comparing them with regulations, detecting inconsistencies and identifying unnecessary administrative procedures.

Used responsibly, the technology could therefore become more than an automation tool; it could serve as a mechanism for auditing institutions and measuring whether existing processes actually create value.

The broader implication is that AI may force society to reconsider why certain jobs and processes exist in the first place.

Some workers will undoubtedly lose valuable careers because of automation, and governments and businesses will need to provide meaningful support during that transition.

But AI could also eliminate work that has become unnecessary, reduce bureaucratic layers and expose inefficiencies that have persisted simply because organizations lacked effective ways to identify them.

Policymakers therefore need to measure both sides of the AI equation: the social and economic costs of displacement as well as the costs of maintaining inefficient systems.

The ultimate goal of technological progress should not simply be to preserve every existing job, but to increase the value society receives from human effort while creating meaningful economic opportunities.

The real test of AI may consequently be less about how many jobs disappear and more about what society does with the productivity that AI creates.

If the technology can eliminate unnecessary work, improve accountability and redirect human talent toward activities that genuinely create value, the AI revolution could become not only a transformation of employment but also an opportunity to redesign inefficient institutions.