The career ladder is disappearing. Are taxes making it worse?

Rishi Sunak is not a man given to alarm. The former British Prime Minister, trained as a Goldman Sachs analyst and steeped in the careful language of finance, has built his post-government career advising two of the most powerful technology companies on earth — Microsoft and Anthropic. When someone with that vantage point describes the current moment as "the most dangerous and the most transformative time in living memory," HR leaders on this side of the Atlantic should pay attention.

Writing this weekend in The Sunday Times, Sunak laid out a stark assessment of what artificial intelligence is doing to the labor market — and what governments and employers must do before the window to respond closes. His argument was directed at British policymakers, but its implications land squarely in American boardrooms.

The numbers are already moving

Sunak's piece points to research by Stanford economist Erik Brynjolfsson showing a 16% decline in hiring for entry-level jobs in the most AI-exposed professions in the United States. That figure, sobering on its own, likely understates what is already underway.

Entry-level hiring at the 15 biggest tech firms fell 25% from 2023 to 2024, according to a report from SignalFire. A Harvard University study tracking 62 million workers across 285,000 US firms found junior positions shrinking at companies integrating AI since 2023, with researchers warning that AI is eroding the bottom rungs of career ladders by automating many of the intellectually mundane tasks that junior employees typically handle.

In the US, entry-level postings in software development and data analysis have plummeted, with some data indicating a 67% decrease in junior tech postings. The effect is not confined to Silicon Valley. Weak hiring has been especially pronounced for knowledge-work roles including marketing and human resources, according to Indeed's Hiring Lab.

The broader labor market is operating in what economists are calling a "low-hire, low-fire" equilibrium. Companies are achieving headcount reduction through attrition — simply choosing not to backfill roles when employees leave. This dynamic hurts entry-level candidates more than anyone else. In a low-hiring environment, the few open roles that do exist are prioritized for immediate impact, with organizations under pressure to maintain margins preferring to hire one experienced senior employee rather than three juniors who require onboarding.

The apprenticeship argument

One of Sunak's most pointed prescriptions for this moment concerns how we train people for work. He argues that apprenticeships — particularly in the professions — become especially valuable in an AI era, because they teach workers how things are actually done, not just the theoretical frameworks that AI systems have already ingested.

It is a view gaining traction in the United States. IEEE Spectrum reports that apprenticeship allows students to learn on the job in a structured program and helps close the experience gap that AI is rapidly widening at entry level — noting that one of the practical skills most likely to be learned on the job is precisely how to use AI for the work itself.

For HR leaders, this reframes the traditional graduate recruitment model in a fundamental way. The assumption that universities produce work-ready talent is eroding fast. The traditional deal of entry-level work — trading rote labor for mentorship — is dying, with the learning curve being automated and leaving early-career professionals stranded between AI agents and senior incumbents.

Organizations that continue to structure their pipelines around the old model will find themselves either unable to attract candidates or unable to develop them once hired.

AI literacy as the new baseline

Sunak frames AI literacy as the contemporary equivalent of a driver's license — the skill you will soon need to do almost any job. The data supports him. According to Stanford's 2025 AI Index report, 78% of organizations are already using AI in at least one part of their work, up from 55% in just one year.

Yet adoption at the organizational level is running far ahead of capability at the individual level. Only about 43% of US workers reported regularly using AI at work last year, and roughly 40% said they were actively disengaged with AI, according to an Indeed Hiring Lab survey.

That gap — between the tools companies are deploying and the skills their employees possess — is one of the most urgent workforce challenges HR professionals face right now. Lifelong learning and upskilling are now a top priority for 75% of US employers. But priority and execution are different things. Most corporate training programs were not designed for the pace and breadth of change that AI demands.

The tax argument, reframed for America

In Britain, Sunak's most politically charged argument concerns the asymmetry in how employers are taxed on human workers versus AI tools. Companies pay tax when they hire a person; they pay nothing comparable when they deploy an AI agent to do the same work. In the US context, the parallel is striking — and the policy trajectory is moving in the opposite direction from the one Sunak recommends.

The recent One Big Beautiful Bill Act permanently reinstated 100% bonus depreciation for qualified capital investments and immediate expensing for R&D expenditures, effectively removing tax friction that once discouraged large-scale investment in technology and automation. US tax policy now actively subsidizes the move from human labor to AI, rewarding companies that reinvest workforce dollars into automation.

The fiscal stakes are significant. In the US, tax on workers accounts for around 84% of federal revenue, consisting of income taxes and payroll taxes. If unemployment is pushed to 20% within five years — as Anthropic CEO Dario Amodei has publicly predicted — employment revenues could be slashed by around a fifth, leaving governments searching for new revenue at precisely the moment when demands on public services would be greatest.

Congress is beginning to stir. Senators Mark Warner and Josh Hawley introduced the AI-Related Job Impact Clarity Act, which would require major employers and federal agencies to report quarterly on AI-related layoffs, new AI-created positions, retrained or reassigned workers, and positions left unfilled due to automation. A bipartisan Economy of the Future Commission Act, introduced in March 2026, would bring together experts in technology, education, and taxation to develop legislative recommendations covering AI workforce training, reskilling, unemployment insurance policy, and taxation. SHRM formally supports the legislation.

The compliance landscape is shifting fast

For HR leaders, the policy conversation is not merely theoretical. New York City's Local Law 144 already requires covered employers to conduct independent bias audits of automated employment decision tools before use and annually thereafter, and to provide advance notice to candidates and employees. Colorado's AI Act, effective June 2026, introduces a risk-based framework classifying employment-related AI systems as high risk and triggering new disclosure and governance obligations.

Critically, employers recruiting remote workers may be subject to these laws even if they are not physically located in the regulating jurisdiction. The patchwork is expanding rapidly, and the organizations best positioned are those that have already begun documenting their AI use in hiring, performance management, and workforce planning.

What HR leaders should do now

Sunak's article is a warning, but it also carries an implicit blueprint. The companies that navigate this transition well will be those that treat it as a workforce strategy problem, not a technology procurement problem.

That means investing seriously in retraining programs rather than treating upskilling as a line item. It means redesigning entry-level roles around the tasks that AI genuinely cannot replicate — judgment, relationships, ethical reasoning, contextual knowledge — rather than simply eliminating those roles. And it means engaging honestly with the emerging regulatory environment rather than waiting for federal clarity that may not come quickly.

Sunak's closing warning bears repeating in any corporate context: the defining question is whether AI ends up automating jobs — producing structural unemployment as a permanent feature of the economy — or augmenting them, helping workers do more and do it better. That choice is not made by governments alone. It is made, every day, by the people responsible for how organizations hire, develop, and deploy their most important asset.

The HR function has never been closer to the center of a question this consequential.