This article is an on-site version of our The AI Shift newsletter. Premium subscribers can sign up here to get the newsletter delivered every Thursday. Standard subscribers can upgrade to Premium here, or explore all FT newsletters
Key Takeaways:
- Bifurcated Labor Impact: New research, including from the US Federal Reserve, consistently demonstrates that AI is slowing employment growth for junior and contractor white-collar roles, particularly in software development, due to its ability to automate “weakly bundled” tasks. This points to a significant restructuring of entry-level professional pipelines across industries.
- Augmentation for Senior Roles: Conversely, AI acts as a powerful assistant for senior professionals with “tightly bundled” and domain-specific expertise, freeing them from lower-value tasks and enhancing overall productivity. This dynamic suggests a widening skills premium and a shift towards higher-value, judgment-intensive work for experienced professionals.
- Strategic Corporate & Policy Imperatives: The deepening understanding of AI’s differential impact necessitates urgent strategic responses from corporations regarding talent management, reskilling, and AI integration. For policymakers, it highlights the need for new educational frameworks and social safety nets to address the evolving nature of work and potential job displacement, with implications for economic stability and growth forecasts.
Welcome back to The AI Shift, our weekly exploration of how AI is reshaping jobs and work. Sarah is away this week, so stepping into her shoes is the FT’s AI editor, Madhumita Murgia. For this edition we are revisiting the big question of whether AI has already started to take white-collar jobs, in light of a flurry of new research and evidence.
John writes
Today’s edition was sparked by a new paper from economists Leland Crane and Paul Soto of the US Federal Reserve, which represents the first time to my knowledge that official labour market statistics have corroborated the story from detailed private payroll data that AI is reducing employment in some pockets of the economy. This is a critical development for economists and market analysts alike, as official data lends significant weight to anecdotal evidence and more granular private studies, providing a clearer signal for macroeconomic forecasting and policy-making.
We have previously reported on research showing a dip in employment for young software developers, based on fine-grained analysis of millions of payroll records, but the finding wasn’t matched by labour force survey data. That gap created uncertainty, making it difficult for investors and corporations to discern the true pace and scale of AI’s impact on the labor market.
That gap has now closed. By using an expanded definition of coders — crucially including the large contingent of contractor software developers as well as coders outside the tech industry — Crane and Soto find a very similar result using the flagship US labour force survey as Stanford’s Erik Brynjolfsson and co-authors did using payrolls. Both estimate that around half a million fewer coders are working today than would have been if pre-LLM-era employment trends had continued. It’s worth noting that this is not an absolute decline in coder employment, but a marked slowdown in growth. For investors, this signals a potential deceleration in labor costs for companies heavily reliant on junior software development, while also highlighting a shift in talent acquisition strategies away from broad-based entry-level hiring towards more specialized or augmented roles.
Just as interesting as the headline findings is the fact that nuances in these results align neatly with several recent papers setting out frameworks for thinking about how AI job displacement is likely to play out and highlighting weaknesses with simple occupation-based or task-based models. These advanced frameworks offer a more granular lens through which market participants can anticipate future labor market shifts and allocate capital more effectively.
A paper last month by LSE professor Luis Garicano and co-authors extends the idea that jobs are bundles of tasks, to consider whether the different activities in a job are a tightly bound bundle or something more akin to an itemised list of discrete activities. In the context of software, a contractor or junior hire generally falls into the latter group: these jobs are weak bundles, with daily work consisting mainly of writing code to spec — tasks that could be given to someone else (or AI) without any disruption to the workflow or the quality of the final product. Here, AI breaks off a large chunk of the job and leaves a role with substantially diminished scope (or obviates the need for that hire or contract). This “unbundling” effect can lead to significant cost efficiencies for companies, potentially boosting margins, but also poses a challenge for staffing agencies and educational institutions focused on entry-level programming skills.
But senior developers, or coders working outside the tech industry in roles where their programming skills are combined with domain-specific expertise, tend to have jobs comprising tightly enmeshed and cross-functional tasks. Here extracting the coding part of the job from all the rest is much harder, so the bundle of tasks remains intact. Instead of becoming a competitor AI becomes an assistant, enhancing rather than eroding the job. This fits with the findings from Brynjolfsson and our own analysis that hiring for senior software roles continues to hold up better than for junior ones. This dynamic reinforces the growing premium on expertise and problem-solving capabilities, suggesting continued wage growth and demand for highly skilled professionals even as entry-level positions face pressure.
The bundling framework is also explored in recent papers by Lukas Freund and Lukas Mann, and by Joshua Gans and Avi Goldfarb, who move beyond the size and interconnectedness of a job’s task bundle to consider the importance of the surviving tasks left after one is automated. When coding is done by AI, senior developers have more time to spend on the many other valuable parts of their job, like translating business needs into product specifications or making judgment calls based on years of accumulated expertise. AI automates a relatively lower-value part of their job and acts as a multiplier on all the rest. But take away coding from a junior developer or contractor and you’re left with very little. In this way, the same technological capability shrinks one job while expanding another — moreover it erodes the junior version of a job even as it enhances the senior version. This “multiplier effect” on senior roles could translate into significant productivity gains for companies, a key factor for long-term economic growth and investor returns.
Between the now-consistent picture on junior coding employment and the expanded framework of jobs as bundles of tasks, it feels to me like we’re developing an increasingly coherent picture of AI job displacement. This clarity is invaluable for investors trying to navigate the “AI boom,” informing decisions on which sectors and companies are best positioned for growth, and which may face structural headwinds in their workforce composition and cost structures.
Madhu writes
This new data comes at an interesting moment, John. OpenAI released a policy blueprint this week that proposes some radical changes to the social contract, in response to what it casts as inevitable job losses and disruption of entire sectors. This move, from a leading AI developer, is not just a PR exercise; it’s a signal to markets and regulators about the scale of anticipated transformation. Of course, it’s in their interest to claim their product will be singularly revolutionary, which can boost valuations and attract further investment, but I’ve also spent the last couple of weeks speaking to investors, analysts and executives from a range of white-collar professions for a piece published today on which jobs are resilient — and which aren’t — in an age of AI agents. The fact that AI is shrinking certain types of employment — mainly early-career jobs — is accepted in these circles, although it is being whispered. The “whispers” indicate a growing awareness of significant, yet often unquantified, shifts in corporate hiring strategies and talent development pipelines, which will eventually manifest in labor market statistics and corporate earnings reports.
The recent research is really interesting for two reasons. First, it confirms what AI companies, white-collar professionals and pretty much any AI user I speak to, are telling me: that AI automation is a double-edged sword. On one hand, AI allows you to supercharge your skills if you are already proficient at your job. One person from a frontier AI company described this as tackling a gnarly project by cloning yourself. This augmentation effect is driving demand for advanced AI tools and platforms, directly benefiting companies like OpenAI, Microsoft, and Google, and boosting their stock performance. On the other hand, if you are just starting out, and don’t yet have the instincts and knowledge developed through hands-on experience, you are more likely to be replaceable. This creates significant challenges for educational institutions and governments seeking to prepare the next generation for the workforce, and for companies managing their entry-level talent pool.
I find it fascinating that this effect seems to be profession-agnostic. I’ve heard it repeated from software engineers, but also journalists, musicians, financial services professionals and lawyers. This universality suggests that the “task bundling” framework could be a powerful predictive tool across various industries, enabling investors to identify sectors and companies most susceptible to labor cost reductions or productivity gains. This is partly because of how the technology works: the errors it makes can often seem random, meaning those without the nous to doubt its outputs are caught out by mistakes more easily. It seems using AI effectively is a skill that only comes with mastery of your subject. This highlights the enduring value of human judgment, critical thinking, and domain expertise, even in an AI-augmented world, influencing where corporations will continue to invest in human capital.
The other thing the research points to is that AI is picking off clusters of tasks that make up a job, starting with the mundane and working its way up the chain to more cognitively demanding ones. The wave of AI agents that we see today, like Anthropic’s Claude Cowork or even OpenAI’s Codex, which use AI to write code, can complete multiple tasks simultaneously, extending the complexity of tasks that can be automated. This accelerating capability of AI agents underscores the urgency for individuals and organizations to adapt, retraining workforces and redesigning job roles to focus on uniquely human skills that AI cannot replicate.
Market Impact:
The converging evidence of AI’s bifurcated impact on the white-collar labor market presents a multifaceted landscape for investors and corporations. Companies that strategically integrate AI to augment their experienced workforce, fostering innovation and efficiency, are likely to see enhanced productivity and potentially superior financial performance. Conversely, those heavily reliant on entry-level or contract roles performing “weakly bundled” tasks face significant pressure to re-evaluate their talent acquisition and development models, potentially leading to workforce reductions or substantial investment in upskilling. This shift will likely drive increased investment in AI-powered tools and platforms, benefiting technology providers, while also spurring demand for specialized consulting in AI implementation and change management. Furthermore, the emerging productivity gains from AI could influence broader economic growth forecasts, potentially impacting interest rate expectations and equity valuations across sectors. Policymakers, confronted with the prospect of structural unemployment at the junior level, will need to consider robust educational reforms and social safety nets to manage the societal transition, adding a layer of regulatory and social risk to the investment calculus for the AI industry.
The discourse around artificial intelligence’s impact on the workforce often oscillates between utopian visions of enhanced productivity and dystopian fears of widespread job displacement. While many early conversations focused on AI’s ability to automate repetitive, low-skill tasks, the evolving capabilities of AI agents are shifting the goalposts dramatically, presenting a more complex challenge to established labor market dynamics. Investors and corporate strategists alike are keenly observing how these advancements translate into real-world efficiency gains, competitive advantage, and ultimately, shareholder value.
Key Takeaways:
- AI Agentic Capabilities Threaten Broad White-Collar Employment: Rapid advancements in AI agents, now handling multi-hour and soon multi-day tasks, signal an impending shift from junior-level job displacement to disrupting experienced white-collar roles, forcing a re-evaluation of human capital strategies across industries.
- Productivity Multiplier vs. Immediate Displacement: While AI’s march is undeniable, current broad job displacement remains limited outside specialized coding roles. This suggests a near-term phase where AI acts more as a productivity enhancer for skilled workers, allowing them to manage complex AI projects and augment creative output.
- Policy & Investment Implications Deepen: The growing discussion around “robot taxes” and public wealth funds, coupled with calls for better data on AI’s employment impact, underscores increasing regulatory scrutiny and the need for proactive corporate and governmental strategies to mitigate economic shocks and guide AI investment.
The core of this evolving narrative lies not just in what AI can do, but in the growing autonomy and ‘agentic’ capabilities of the technology—a development poised to redefine productivity benchmarks and operational expenditure for enterprises. This shift fundamentally alters the calculus for human capital allocation and workforce planning.
On these topics, I recently sat down with Mark Chen, head of research at OpenAI, a company at the forefront of AI innovation and a bellwether for venture capital flows into the sector. Chen highlighted the METR benchmark, a critical metric measuring AI agent performance by the complexity and duration of tasks they can independently complete. This metric, closely watched by investors for its implications on future productivity gains, has shown an exponential increase over recent months. Chen observed that a mere year ago, AI models were tackling tasks measurable in minutes – processes typically assigned to entry-level or junior staff. “Now we’re dealing with tasks in the hours,” he explained, “and if you extrapolate that forward, we’re soon going to be having our models do tasks reliably that would take humans days.” He cited the example of an AI autonomously developing a functional piece of software, a capability that could profoundly reshape software development lifecycles and associated labor costs.
This rapid evolution strongly suggests that while early job displacement has disproportionately affected junior employees—often in roles easily automated—the scope is broadening. The advent of AI agents capable of higher-level, multi-day tasks implies that even experienced professionals in fields like finance, consulting, and advanced engineering will increasingly transition from primary creative roles to overseeing sophisticated AI-driven projects, fundamentally altering the value proposition of human capital. This raises critical questions for corporate strategy: how to reskill, how to restructure, and how to maintain a competitive edge in a rapidly automating landscape. What are your thoughts on this disruptive trajectory, John?
John responds
Thanks Madhu, it’s indeed compelling to see these industry whispers—often reflected in enterprise technology spending reports and venture capital allocations—aligning with the hard data on AI’s accelerating capabilities. There’s undeniably no sign that the forward march of AI’s functional scope is decelerating, a trend that will, unequivocally, lead to significant job disruption across various economic sectors. However, my outlook, while acknowledging this transformative force, remains optimistic on two key fronts. First, I believe there are structural reasons why we are still observing relatively limited, widespread job displacement beyond specialized coding roles—a phenomenon I expect to persist for a while. This ‘last mile problem’ for AI often involves the necessity for human judgment in ambiguous situations, navigating complex regulatory frameworks, or ensuring ethical oversight, providing a critical buffer. Companies are still grappling with integrating these advanced AI agents into existing workflows, requiring human intervention for validation, customization, and error correction. This translates into a strategic imperative for businesses to invest in upskilling their workforce rather than simply downsizing, at least in the near-to-medium term, to maximize their return on AI investments.
Second, and perhaps more personally impactful, as someone who has increasingly transitioned into managing agentic AI projects, I’ve experienced firsthand how AI can serve as a profound multiplier of human creativity and strategic thinking, rather than a direct replacement. It offloads the tedious, iterative, or data-intensive aspects, freeing up human intellect for higher-order problem-solving, innovation, and strategic direction. This redefines the human-AI collaboration paradigm, suggesting a future where the premium is placed on cognitive agility and the ability to effectively ‘prompt engineer’ and direct sophisticated AI tools for competitive advantage. For investors, this implies a focus on companies that not only develop cutting-edge AI but also demonstrate a clear strategy for integrating these tools to achieve measurable productivity gains and foster innovation.
Recommended reading
OpenAI’s version of a New Deal is a 13-page document proposing robot taxes and a public wealth fund, among other things, to protect from the economic shocks of artificial intelligence (Madhu)
Over at MIT Technology Review, Alex Imas makes the case to James O’Donnell that we could have a much clearer picture on how AI is changing employment if we had more and better data (John)
Market Impact:
The implications of these accelerating AI capabilities on global markets are profound and multi-faceted. From an investment perspective, we anticipate continued robust capital flows into AI development and integration, favoring companies that can demonstrate clear pathways to operational efficiencies, cost reductions, and novel revenue streams through AI adoption. This could translate into a re-rating of sectors historically reliant on extensive human capital, such as professional services, finance, and software development, as AI drives down labor costs while potentially boosting output quality. However, this also introduces significant market risks: widespread job displacement, if not managed with proactive reskilling and social safety nets, could depress consumer spending, leading to broader economic contraction. Governments and regulators are likely to face increasing pressure to intervene, with proposals like “robot taxes” or public wealth funds, as suggested by OpenAI, gaining traction. Such policies, while aiming to mitigate societal disruption, could also introduce new layers of complexity and cost for businesses. Companies able to effectively navigate this evolving landscape—investing strategically in AI, managing human capital transitions, and adapting to potential regulatory shifts—will likely emerge as market leaders, while those that fail to adapt risk significant competitive disadvantage and erosion of shareholder value. The next decade will undoubtedly be defined by how effectively economies harness AI’s transformative power while addressing its inherent challenges.