The Last of Us (White-Collar Workers)

As white-collar professionals contemplate their future, they might want to pick up a copy of Nick Bostrom’s book Superintelligence: Paths, Dangers, Strategies, which examines the threats posed by artificial intelligence. In it, Bostrom draws on a compelling historical parallel: “When horses became obsolete as a source of moveable power, many were sold off to meatpackers to be processed into dog food, bone meal, leather, and glue. These animals had no alternative employment through which to earn their keep. In the United States, there were about 26 million horses in 1915. By the early 1950s, 2 million remained.”

Like it or not, raising the fate of former workhorses when discussing the outlook for today’s business professionals is not completely an exercise in hyperbole. While slaughterhouses may not be in our future, the obsolescence of white-collar workers is the openly stated expectation of many leaders building the AI platforms set to be adopted across nearly every law firm, consulting shop, accounting practice, and investment bank.

This article highlights how the conversation about AI’s impact on white-collar work needs to move beyond the theoretical. The question is no longer about whether AI will disrupt professional services, but when. Drawing on what is already unfolding in the market, this article examines the implications for the last of us performing traditional white-collar work.

Warning Signs and Dire Predictions

Earlier this year, Baker McKenzie, one of the world’s largest law firms, announced the elimination of up to 1,000 support positions, reportedly attributing the cuts, at least in part, to its growing use of AI. Within days, commentators debated whether the layoffs represented a genuine technological inflection point or simply another case of so-called “AI washing,” where companies invoke the language of innovation to justify cost-cutting that would have happened regardless.

As Above the Law contributor Joe Patrice put it, “I don’t know if Baker McKenzie really believes it can replace 700 staff with AI or if blaming the bots just provides a convenient excuse for general management missteps.” Either way, when it comes to cutting headcount, Baker McKenzie is not alone.

Last year, when over 100,000 employees were reportedly impacted by AI-driven layoffs, the Big Four accounting firms—Deloitte, EY, KPMG, and PwC—announced significant restructurings while McKinsey reportedly began planning reductions of up to 10 per cent of its non-client-facing workforce. This year, Accenture, Amazon, Citigroup, Dell, Intel, Microsoft, TCS, and UPS each announced AI-related layoffs affecting more than 10,000 employees, according to a third-party review of company disclosures, earnings calls, and media reports. In the first half of 2026 alone, almost 80,000 employees across more than 45 companies were reportedly impacted by AI-driven layoffs. In Canada, layoff trackers report that the tech sector has shed thousands of jobs over the past eighteen months, with cuts at companies such as Shopify and OpenText.

These are not fringe startups experimenting with automation. They are the institutions that define white-collar professional life. And they appear to be simultaneously redrawing the terms of human employment.

How bad could it get? In February 2026, Mustafa Suleyman, the CEO of Microsoft AI, sat down with The Financial Times and outlined his company’s expectations for the technology, which include “human-level performance on most, if not all, professional tasks” within 12 to 18 months. After Suleyman named lawyers, accountants, project managers, and marketing professionals as immediately vulnerable, a Futurism headline announced “Microsoft CEO: Virtually All White-Collar Tasks Will Be Automated Within a Year and a Half.”

The Futurism headline ran under an “Officepocalypse” label, but the article itself was skeptical. When reporting on Suleyman’s comments, Futurism contributor Frank Landymore argued that the current layoff numbers are misleading because AI washing is boosting the perceived impact of AI adoption on employment. He also noted that it remains unclear how sustainable it will be for companies making real AI-related cuts “to heavily depend on the tech in the long-term.”

 Nevertheless, a few months earlier, the headlines were about Anthropic CEO Dario Amodei sounding the alarm over a potential “white-collar bloodbath.” In an interview with Axios, Amodei warned that AI could eliminate half of all entry-level white-collar jobs within five years, potentially spiking US unemployment to 20 per cent. “We, as the producers of this technology, have a duty and an obligation to be honest about what is coming,” he insisted.

Some question whether the industry’s dire predictions are really just sophisticated marketing that aims to inflate the perceived power of AI products. And maybe it is all self-serving scaremongering. But dismissing the predictions entirely, just because the messengers have a commercial interest, could prove to be a serious and myopic mistake.

The Changing Economics of Knowledge

Professional services have historically operated in a market in which clients paid for specialized human judgment because there was no real substitute. For decades, this market was defined by the scarcity of knowledge. Because expertise was expensive to acquire, slow to develop, and hard to replicate, significant information asymmetry existed between professional services providers and clients. And because supply was constrained while demand was strong and barriers to entry high, firms that aggregated and deployed expertise—whether in law, consulting, accounting, or finance—could traditionally command premium fees.

But what has long been a seller’s market is starting to rebalance because (a) access to knowledge is no longer gated by expensive institutional infrastructure and (b) AI is driving the marginal cost of knowledge work closer and closer to zero.

The first draft of a contract, the first pass of a due diligence review, the initial synthesis of a research question. All of these tasks once required hours and hours of billable professional time, but AI can perform them in minutes at negligible incremental cost. As a result, some clients will choose to accomplish much of these tasks internally. Others will simply demand lower prices. After all, when the marginal cost of a service falls while demand holds, prices eventually follow. Put simply, when a large language model can synthesize case law, draft contract provisions, build financial models, prepare tax memoranda, or produce a strategic market analysis in a fraction of the time it would take a junior professional, the professional services market’s traditional pricing structure breaks.

In law, the billable hour has persisted for generations, but not because it accurately reflects the value of legal work. As Stephanie Corey and Ken Callander noted in Above the Law, the billable hour model became standard because it absorbed uncertainty. When scope was unclear and outcomes were hard to predict, time functioned as a hedge. But the logic behind billable hours is being disrupted. As Corey and Callander explained, “When intelligent systems can complete discrete legal tasks faster, more consistently, and at lower cost, time stops working as a credible proxy for value.”

This is no longer speculative. Tools like Harvey, Microsoft Copilot, and Legora are already moving into Canada’s top firms, with several of the largest now deploying them firmwide rather than piloting them in isolated groups. Nor is this confined to law. Models now draft spreadsheets, build financial analyses, and assemble first-draft audits across accounting, consulting, and banking. Of course, outputs are not yet flawless—inaccuracies persist, and the workflows around the technology remain inefficient in places. But this is a nascent technology, and such technologies improve while firms “restructure” around these newfound efficiencies.

Keep in mind that legal clients have been paying for inputs (i.e., hours, bodies, effort) when what they actually value is the output (i.e., the legal opinion, the strategic recommendation, the clean audit). And when a review of commercial agreements that once required 20 to 40 associate hours can be completed in five with the same outcome delivered more predictably, is it really worth the same as it was before? Or does traditional billing result in a mispricing?

As Above the Law contributor Stephen Embry put it, “This disconnect exposes an uncomfortable truth for firms: future legal work will require less time with fewer lawyers and, conceivably, less revenue. It creates a gap that can’t be closed simply by raising rates and cutting expenses. Instead, it may require a different business model altogether.”

As things stand, the legal profession isn’t feeling much pressure to evolve thanks to strong demand that has allowed worked rates to set new records, countering a decline in billable hours per lawyer that has challenged profits in recent years. In 2025, Kirkland & Ellis LLP became the first firm to surpass US$10 billion in annual revenue. With regulatory shifts and general economic chaos helping produce a banner year for demand growth last year, average US law firm profit growth increased 13 per cent, according to the 2026 Report on the State of the US Legal Market, published by Thomson Reuters and the Center on Ethics and the Legal Profession at Georgetown Law.

“The pyramid model, in which partners supervise teams of associates billing thousands of hours, has served as the economic engine of professional services firms. And its cracking creates a tension that the industry has not yet resolved, nor priced in.”

But with soaring profits, perhaps the legal industry has been lulled into ignoring warning signs about the sustainability of the model delivering them. That is what makes this moment feel so fragile. The hour-driven architecture that has defined Big Law for half a century by equating time with value has never performed better, which is precisely why so few firms feel compelled to question, at least for now, whether it is fast approaching its best-before date as AI-driven efficiency gains take root.

In consulting, the talk of mass layoffs at McKinsey has at least spawned a sense of urgency. According to Nitin Seth, a former director of McKinsey’s Global Knowledge Centre, the cuts signal an irreversible shift in how the industry creates value. As he notes in a Fast Company commentary, “Firms like McKinsey built a powerful competitive moat by hiring the best analytical minds from top universities—excelling at data synthesis, first-principles problem-solving, and translating insight into recommendations. In the AI age, however, that advantage is becoming commoditized.”

In July 2023, McKinsey deployed a proprietary AI chatbot that was widely adopted, with over 70 per cent of the firm now active on the platform, which reportedly delivers up to 30 per cent time savings in searching for and synthesizing knowledge. According to observers, the fundamental insight here is that AI isn’t helping junior consultants—it is replacing them, at least in their current core function as data-processing “workhorses.” As AI project manager Roberto Carreras writes in a Medium article, McKinsey’s employee headcount is “already 25% below its historical peak. This shift is not accidental. The evidence lies in the massive adoption of internal intelligent agents. When a tool can reduce research and synthesis time by 30%, the question for a McKinsey partner becomes inevitable: Why do I need ten analysts if three plus an AI can deliver the same result in less time?”

Carreras and others insist we are seeing the industry’s rigid pyramid structure, which was built on a base of overworked junior analysts, start to fracture. And what this means for entry-level professional workers is perhaps the most consequential, and least discussed, unknown.

For decades, professional firms have operated on a well-understood apprenticeship pipeline. Junior lawyers have reviewed documents, drafted initial memoranda, and conducted legal research. Junior consultants have built presentation slide decks, run models, and synthesized data. Junior accountants have prepared statements, reconciled ledgers, and tested controls. These are not glamorous tasks, but they have served a vital developmental function as the mechanism by which professionals develop judgment and eventually earn the trust required to advise clients independently. And if AI takes over the grunt work, how will firms train the next generation of senior professionals?

The disruption has already reached the job titles themselves. Deloitte, which like McKinsey and other consultancies has invested heavily in AI, recently started to retire the traditional analyst–consultant–manager ladder across its US workforce. According to media reports, the move represents a divergence from the familiar progression track in favour of more specialized, function-specific designations.

The pyramid model, in which partners supervise teams of associates billing thousands of hours, has served as the economic engine of professional services firms. And its cracking creates a tension that the industry has not yet resolved, nor priced in. On one hand, junior professionals may be given more responsibility early on. With AI handling routine work, for example, a second-year associate might be expected to exercise judgment and client-facing skills that have traditionally been reserved for those several years more senior. That could be genuinely empowering. On the other hand, there may simply be fewer juniors, as firms thin the bottom of the pyramid that once propped the entire model up.

Either way, the billing math changes as firms need fewer people at every level, or fundamentally different people whose value lies in exercising contextual judgment that AI cannot yet replicate. The risk here is not simply economic. It is generational. The class of 2030 may graduate into a professional job market that looks nothing like the one their predecessors entered. The traditional gateway roles that allowed bright, ambitious graduates to break into professional services are narrowing quickly.

As things stand, it seems possible, maybe even likely, that in a few years’ time, a new associate will no longer be able to count on securing promotion and demonstrating value through sheer volume of work on routine matters. What comes next is not yet visible, but the traditional pathways seem to be quietly disappearing.

The Uncertain Road Ahead

AI could develop more slowly than its creators expect. Adoption may lag capability by years. Entirely new categories of work may emerge. All of these outcomes are possible. Regulatory intervention could also slow societal disruption and worker displacement. That said, it clearly seems to be time, as Amodei advises, to “stop sugar-coating” the threat that AI poses to the professional services market, which runs deeper than headcount. After all, while the specifics vary, market developments and industry predictions suggest that the capabilities of AI systems are advancing faster than most professionals, regulators, and educators seem to appreciate.

So, while it is impossible to know how things will play out, the following scenarios can’t be ruled out:

Near-Term (2026–2028). Firms intensify experimentation and selective deployment of AI across workflows. Displacement remains concentrated in support staff roles, particularly in research, document review, and basic analysis. Hiring slows and incoming class sizes shrink. Promotion timelines may also begin to stretch. At the same time, revenue will likely rise, as efficiency gains and pricing inertia allow firms to maintain or even expand margins despite tightening labour markets.

Medium-Term (2029–2031). Structural change becomes increasingly visible and difficult to reverse. Pricing models begin shifting more decisively toward value-based and fixed-fee arrangements, particularly for standardized work. The traditional pyramid flattens as the economic logic of large junior cohorts weakens, replaced by smaller teams led by more senior professionals acting as “orchestrators” of AI systems. Headcount declines across law, consulting, accounting, and finance, not through abrupt shocks but through sustained reductions in hiring and gradual workforce compression. Consolidation accelerates. Firms that invested early and captured the productivity gains begin leveraging excess capital to outpace slower rivals, undercutting on price where work is commoditized, absorbing smaller practices, and reinvesting the surplus into faster growth. Scale begets scale, and the gap between the firms that adapted and those that hesitated widens into something harder to close. At the same time, productivity per professional rises significantly, reshaping expectations around output, leverage, and career progression.

Long-Term (2032–2036). Outcomes begin to diverge more sharply. In a moderate scenario, firms evolve into lean, highly integrated organizations in which AI is embedded across all functions, resulting in a small but substantially more productive workforce. Professional roles persist but are redefined around judgment, client management, and complex problem framing. In a more disruptive scenario, continued advances toward artificial general intelligence extend automation into higher-order cognitive work, including tasks previously thought to require deep expertise and experience. Under these conditions, the traditional professional services model will likely face fundamental pressure, raising questions not just about firm structure but about the long-term role of human professionals within these industries.

The truth of what to expect, as is often the case, probably sits somewhere between the worst-case and best-case scenarios. But even if current progress stalls, what used to feel like a distant possibility is starting to look like an imminent reality. The Baker McKenzie and McKinsey reductions in back-office roles, the Big Four’s pivot toward AI-integrated staffing, the erosion of the billable hour, the industry warnings. Taken individually, each could be explained away. Taken together, they suggest that professional services may be approaching a structural inflection point.

For firms, failing to prepare for this, even provisionally, would be a mistake. That means rethinking the current business model, which looks increasingly fragile. The good news for individual professionals, especially those early in their careers, is that the value of domain expertise, judgment, and the ability to navigate complex human relationships is not diminishing. If anything, these capabilities will become more valuable as routine work gets automated. That said, relying on the traditional career playbook may no longer prove a guaranteed path to long-term security.

For society at large, the central question is whether we are prepared for the possibility, however uncertain, that a meaningful share of the professional workforce could be displaced within years rather than decades. At present, the answer seems to be no.

Unlike manufacturing workers in prior eras of automation, professional workers lack strong union representation and tend toward individual rather than collective responses. As a result, AI-related layoffs have generated little coordinated resistance. That could change. As Konrad Yakabuski recently reported in The Globe and Mail, a loose coalition of Hamilton residents, environmentalists, and anti-capitalist activists successfully defeated a proposed data centre project on Steeltown’s waterfront on the same day Canadian AI Minister Evan Solomon tabled Ottawa’s “bullish vision to transform Canada from AI laggard to leader.”

Nevertheless, with social pushback scattered, the impact of AI on white-collar workers will likely depend on how the transition is managed by politicians. And governments have largely been ignoring the workforce-related implications of AI. In the United States, the current political environment favours deregulation and market-driven adjustment. In Europe, the AI Act establishes a risk-based regulatory framework, but it focuses mainly on safety and transparency rather than on the labour market.

Meanwhile, AI leaders are suggesting unrealistic utopian solutions to the disruption they expect to drive. A recent OpenAI policy paper called on society to respond to technological upheaval with an “ambitious form of industrial policy, one that reflects the ability of democratic societies to act collectively, at scale, to shape their economic future so that superintelligence benefits everyone.” Amongst other challenging things, this would involve “reimagining the social contract, mediating between capital and labor, and encouraging broad distribution of the benefits of technological progress while preserving pluralism, constitutional checks and balances, and freedom to innovate.”

OpenAI’s Sam Altman previously suggested industry profits could fund a universal basic income system. “People would be entrusted,” he explained, “to use the money however they needed or wanted—for better education, healthcare, housing, starting a company, whatever…. Poverty would be greatly reduced, and many more people would have a shot at the life they want.”

Not to be outdone, Elon Musk talks about a future in which everyone can have “any goods or services that they want” thanks to a “universal high income.” AI leaders, of course, offer no guarantees of good times ahead. Musk counters his vision of a utopian future with Terminator scenarios, while OpenAI’s policy paper openly admits the future could be one in which “most people lack agency and access to AI-driven opportunity.”

Critics, of course, question both the feasibility and desirability of trying to counter AI disruption with universal basic income schemes. According to venture capitalist Marc Andreessen, universal basic income would turn people into unproductive “zoo animals to be farmed by the state.” Meanwhile, OpenAI’s own three-year universal basic income experiment provided US$1,000 per month to low-income participants and produced mixed results. But even if a universal basic income system is in the cards, it is almost certainly not the whole answer, and it clearly won’t happen in time to mitigate the impact on professional services if the industry predictions prove even remotely close to current estimates.

Bostrom’s horses were displaced not because they had done anything wrong but because the economy stopped needing them. Today’s white-collar professionals should at least entertain the possibility that they face a version of the same fate. The market appears headed for a structural shift that neither the public nor the private sector seems prepared to absorb. Those who recognize it early will have a chance to adapt before the most dire predictions arrive. The rest may not.

Disclaimer: This article reflects the author’s personal views and does not represent the position of any employer, institution, or professional organization. It is intended as a speculative and analytical contribution to an important public conversation and should not be construed as legal, financial, or career advice.