Anthropic's AI is Firing Coders

Anthropic’s recent research on AI’s labor market impacts upends conventional wisdom about which jobs artificial intelligence threatens. Their "Labor Market Impacts of AI" report, published in March 2026, introduced observed exposure, a novel metric fusing large language model capabilities with real-world AI usage data. This analysis revealed highly skilled occupations, not just low-skill roles, face the most significant impact.

Among the professions most exposed are computer programmers, customer service representatives, and financial analysts. These are often educated, specialized roles, directly contradicting the long-held belief that automation primarily displaces manual or repetitive tasks. The data suggests AI's reach extends far into the knowledge economy.

While widespread unemployment has not yet materialized, the early tremors are palpable in entry-level hiring. Anthropic's research found no systematic increase in unemployment rates for exposed workers since late 2022. However, hiring for workers aged 22-25 in these vulnerable fields has already decelerated by approximately 14% compared to 2022 job-finding rates.

Long-term projections from the Bureau of Labor Statistics (BLS) paint a clearer picture of future shifts. They forecast a tangible slowdown in job growth, predicting a 0.6 percentage point drop for every 10% increase in AI exposure through 2034. This indicates a structural transformation, not merely a transient economic ripple.

Anthropic offers a striking glimpse into AI's direct impact on development. Engineers at the company now ship eight times more code per quarter than in previous periods, a dramatic increase in productivity. Claude, their proprietary AI, writes over 80% of the production code merged into Anthropic’s main codebase, showcasing a live example of AI performing core development tasks.

This isn't merely a static efficiency gain; the pace of AI development is accelerating. The complexity and length of tasks AI can reliably complete autonomously now double approximately every four months. This represents a significant acceleration from the previous trend, which saw task length doubling every seven months.

Such rapid advancements point toward a future dominated by Recursive Self-Improvement (RSI). In this paradigm, AI systems will increasingly manage their own development cycles, delegating complex tasks to other AIs with only high-level human oversight. This shift could redefine the very nature of software engineering, transforming human roles from hands-on coding to strategic direction.

AI adoption initiates a profound paradox: individual roles face automation, yet companies leveraging AI often experience explosive growth. Anthropic’s engineers now ship 8 times more code per quarter compared to 2021-2025 levels, with Claude writing over 80% of their production code. This unprecedented productivity gain does not necessarily lead to net job losses across the board; it can paradoxically fuel overall company expansion and even create new, albeit transformed, roles within previously "high-exposure" sectors.

Crucially, the very nature of these jobs shifts fundamentally. The skills required are evolving more than twice as fast as before, demanding uniquely human capabilities that AI, despite its prowess, cannot yet replicate. Future success hinges on mastering abilities such as: - Judgment in ambiguous situations - Creativity for novel problem-solving - Empathy for complex human interactions

This dynamic is rapidly forging a distinct "two-track" labor market. On one track, workers who proactively adapt to collaborate with sophisticated AI systems will likely experience accelerated career trajectories and enhanced wage growth. On the other, those unwilling or unable to integrate AI into their daily workflows risk rapid obsolescence, as their tasks become fully automated. Understanding this shift is critical for navigating the future of work, especially as AI systems increasingly take on their own development, as explored in Anthropic’s insights on advanced AI development. When AI builds itself - Anthropic

The rapid ascent of AI demands a career pivot, not merely tool adoption. Instead of frantically learning every new AI interface, individuals must master first-principles thinking. AI increasingly automates the 'how'—the execution of tasks—making deep understanding of domain fundamentals, rather than superficial tool proficiency, the enduring asset in any field.

Cultivate uniquely human skills. While AI systems like Claude now write over 80% of Anthropic's production code, they cannot yet replicate genuine strategic thinking, complex problem-solving, or emotional intelligence. These capabilities are becoming premium differentiators, indispensable in a landscape where routine cognitive tasks across professions are increasingly delegated to machines.

This moment is not for defense, but for strategic reinvention. Anthropic itself, a leader in AI development, committed €170M to studying AI's labor market disruption—a clear signal of the massive, systemic change ahead. Embrace this shift by evolving your core value proposition beyond automation, ensuring your expertise remains irreplaceable even as AI reconfigures entire industries.

What is recursive self-improvement (RSI) in AI?

Recursive self-improvement is when an AI system can autonomously improve its own code and capabilities without human intervention, potentially leading to exponential growth in its intelligence and performance.

Which jobs are most at risk from AI according to Anthropic?

Contrary to common belief, Anthropic's research shows skilled, educated roles like computer programmers, financial analysts, and customer service representatives are among the most exposed based on real-world AI usage.

Is AI causing widespread unemployment right now?

Anthropic's study found no systematic increase in unemployment in exposed fields yet. However, it did find tentative evidence that the hiring rate for younger workers (22-25) in these roles has dropped by about 14% since 2022.

How is Anthropic using AI to build AI?

Anthropic's engineers are now 8 times more productive, with over 80% of the code merged into their main codebase being written by their own AI, Claude. This is a prime example of AI accelerating its own development.