AI, Automation Risk, and the Future of Work

Artificial intelligence is rapidly shifting from an emerging technology to a core driver of productivity across the global economy. Nearly 80% of U.S. companies already report using AI in some capacity, yet close to 90% indicate that its impact on business operations has so far been limited. This apparent contradiction reflects where the technology currently sits in its adoption cycle: deployment is widespread, but deep organizational transformation is still in its early stages.

The labor market is therefore entering a transitional period rather than experiencing immediate disruption. While headlines frequently emphasize mass job displacement, the evidence suggests a more nuanced reality. AI is changing the composition of work faster than it is reducing overall employment. The occupations facing the highest levels of automation risk are often those with the strongest ability to evolve alongside AI, suggesting that augmentation—not replacement—will define much of the next decade.

The first chart illustrates this relationship by comparing occupations according to two dimensions: risk of displacement from AI and adaptive capacity. The results reveal an important pattern. Research finds roughly a 60% correlation between automation risk and adaptability, indicating that occupations most exposed to AI frequently possess the skills necessary to benefit from technological change rather than simply be eliminated.

This distinction is particularly visible among knowledge-intensive professions. Software developers, financial and investment analysts, aerospace engineers, lawyers, accountants, and registered nurses all exhibit relatively high automation exposure while simultaneously demonstrating strong adaptive capacity. These occupations involve substantial amounts of information processing, documentation, and analysis—tasks increasingly supported by AI—but they also depend heavily on judgment, interpretation, communication, and decision-making that remain difficult to automate.

For software developers, AI coding assistants increasingly generate routine code, identify bugs, and automate testing. Rather than reducing demand for engineers, these tools shift their focus toward architecture, systems integration, product design, and higher-value problem solving. Similarly, financial analysts can automate portions of financial modeling, research, and data gathering, allowing greater attention to investment judgment, client advisory work, and strategic decision-making.

Healthcare presents another example of AI augmenting rather than replacing professionals. Registered nurses can leverage AI for documentation, patient monitoring, scheduling, and administrative tasks, enabling more time for direct patient care—activities that continue to require human empathy, communication, and clinical judgment.

The opposite pattern appears among occupations such as secretaries, administrative assistants, customer service representatives, and retail sales personnel. These roles combine relatively high automation exposure with significantly lower adaptive capacity, making them more vulnerable to structural workforce changes. Because a larger proportion of their daily responsibilities consists of repetitive, standardized, and rules-based tasks, AI systems can automate meaningful portions of their workflows with relatively limited human intervention.

However, even in these cases, complete job replacement remains unlikely in the near term. More commonly, organizations are redesigning roles by automating routine activities while reallocating employees toward customer relationships, exception handling, quality assurance, and complex problem resolution.

Importantly, bubble size in the chart represents employment growth since 2000. Many of today's fastest-growing occupations are also among the most exposed to AI. Rather than indicating future collapse, this reflects the fact that rapidly growing knowledge occupations often consist of digital, information-based tasks that AI can augment effectively.

While automation risk provides a theoretical framework, observed AI adoption offers a more practical view of where AI is already changing work today.

Anthropic's observed exposure measure combines theoretical AI capability with actual usage data, emphasizing work-related and automated applications rather than simple experimentation. The findings show that AI remains well below its theoretical potential. Although large language models could theoretically automate far more work, real-world implementation still covers only a fraction of eligible tasks.

Among occupations currently exhibiting the greatest observed exposure, computer programmers rank first, with approximately 74.5% of core tasks already exposed to AI-assisted automation. These tasks include writing, updating, and maintaining software programs, reflecting the rapid adoption of coding assistants across software development teams.

Customer service representatives follow closely, with 70.1% observed exposure as AI-powered chatbots, voice agents, and workflow automation increasingly handle customer inquiries, order processing, and routine support interactions. Data entry specialists, medical record professionals, and market research analysts similarly rank among the most exposed occupations because much of their work centers on structured information processing that AI performs efficiently.

Financial and investment analysts also appear prominently, with over 57% observed exposure. Rather than replacing analysts, AI increasingly automates data gathering, financial screening, forecasting support, and report generation, enabling professionals to focus on investment judgment, portfolio strategy, and client advisory services.

Notably, exposure does not automatically translate into unemployment. Recent labor market evidence finds no statistically significant increase in unemployment among highly exposed occupations since generative AI entered mainstream adoption in late 2022. Instead, the earliest measurable effect appears in hiring. Employment among entry-level workers in highly exposed occupations has been approximately 13% lower than comparable less-exposed roles since 2022, while hiring rates among workers aged 22 to 25 have begun to soften. These findings suggest that employers are becoming more selective in adding junior positions as AI absorbs routine entry-level work.

Looking ahead, demographics may ultimately prove more important than automation itself. U.S. working-age population growth is expected to slow sharply over the coming decade, with annual employment growth projected at only 0.5%. In this environment, AI-driven productivity becomes essential for sustaining economic growth. CBRE forecasts productivity growth accelerating from an average of 1.4% over the past 25 years to 2.1% annually through 2030, largely driven by AI adoption.

History suggests that transformative technologies rarely reduce employment permanently. Instead, they reshape occupations while creating entirely new categories of work. Approximately 20% of today's jobs did not exist in 1999, underscoring how technological innovation continuously redefines labor markets. AI appears poised to follow a similar trajectory. The greatest impact is likely to be on how work is performed rather than whether work exists, making workforce adaptability, continuous learning, and AI fluency increasingly valuable competitive advantages for both employees and employers.

Sources & References

Anthropic. (2026). Labor market impacts of AI: A new measure and early evidence. https://www.anthropic.com/research/labor-market-impacts 

CBRE. (2026). AI’s Impact on the Economy, Employment & Productivity. https://www.cbre.com/insights/reports/ais-impact-on-the-economy-employment-and-productivity