A groundbreaking new study led by Enrico Maria Fenoaltea and colleagues sheds light on the complex and nuanced ways artificial intelligence (AI) technologies, especially those developed by startups, are poised to transform the labor market. Unlike previous analyses that largely focused on the theoretical capabilities of large language models (LLMs) and other AI platforms, this research leverages real-world data on AI startups funded by venture capital to map likely near-term occupational impacts. By validating and harnessing a version of Meta’s Llama3 LLM and cross-referencing AI products with detailed occupational task descriptions from the O*NET database, the study introduces a novel framework to assess which jobs may face significant AI exposure and which remain resistant to automation in practice.
This approach highlights that the AI revolution will not unfold uniformly across all sectors or jobs. Instead, it uncovers a mosaic in which AI adoption is influenced as much by social and economic factors as by pure technical feasibility. The resulting Occupational AI Startup Exposure (AISE) index attempts to quantify this potential by evaluating not just AI’s ability to perform specific tasks, but by incorporating nuanced insights from investor funding patterns that signal which AI applications are viable and scalable in the marketplace. Thus, the AISE stands as a pragmatic and dynamic indicator of the future landscape of AI-driven change in employment.
Central to the findings is the observation that certain occupational categories, notably those involving structured, data-rich office work, show high AI exposure under the AISE framework. Office clerks, data scientists, and roles in information systems management emerge as professions likely to undergo significant transformation, as AI is poised either to complement or substitute many of their routine and analytical functions. Similarly, marketing specialists and market research analysts are highlighted, reflecting AI’s growing role in data analytics, customer segmentation, and predictive modeling that drives modern business strategies.
Conversely, occupations grounded in manual labor and hands-on skills, such as chefs, construction workers, and athletes, register markedly lower AISE scores. These roles require fine motor skills, creativity, situational adaptability, and physical presence—attributes AI and automation technologies currently find challenging to replicate. The study provides compelling evidence that the physicality and social-personal elements embedded in these occupations act as a natural shield against immediate AI disruption, underscoring the limits of AI’s practical reach despite theoretical capabilities.
A particularly striking insight from the study is the disparity between theoretical AI abilities and real-world adoption as shaped by societal acceptance, trust, and ethical considerations. For instance, although LLMs could hypothetically perform tasks associated with high-responsibility professions such as high school teaching, judicial decision-making, or counseling, the researchers note that human reluctance to entrust sensitive roles involving ethical judgments, social nuance, and personal empathy curtails near-term AI integration in these areas. This finding challenges overly deterministic views of AI’s impact and stresses the social context as a major determinant of AI utility and adoption.
The research highlights that the AISE index predicts significantly lower exposure for occupations demanding advanced education (master’s degrees or higher) and substantial professional experience. This suggests that cognitive tasks combined with complex ethical reasoning and domain-specific expertise remain more resilient, at least for the foreseeable future, to AI-driven substitution. With this nuanced understanding, policymakers, businesses, and workers can better anticipate which skills and roles will thrive and which might require adaptation in an AI-influenced labor market.
Investors’ focus on economically viable and societally acceptable AI applications further channels which innovations reach market maturity. Funded AI startups tend to prioritize scalable solutions that solve concrete problems and generate clear returns, distinguishing this market-driven phenomenon from more speculative AI potentials. This dynamic feedback loop—where societal needs, investor confidence, and technological feasibility interact—ultimately shapes the trajectory and uneven diffusion of AI across industries and roles.
Taking a macroscopic view, the study envisions AI integration as a gradual and selective wave rather than an indiscriminate technological upheaval. The evidence contradicts sensationalist narratives about widespread displacement and instead reveals AI’s trajectory as shaped by a complex interplay of technical competencies, regulatory landscapes, social trust, and economic incentives. This diffusion pattern is likely to yield a differentiated labor market where certain sectors leap forward in productivity, while others evolve more cautiously or preserve human-centric dominance.
Importantly, this research methodology—leveraging detailed AI product information, venture funding data, and occupational task analysis—represents a pioneering attempt to ground AI impact assessments within observable market trends rather than abstract theoretical predictions alone. Such grounded insights can help stakeholders better prepare for AI-driven transformations by focusing on sectors where AI investments and product development actively signal occupational change. This approach bridges a critical gap between AI’s technical potential and its social and economic realities.
Moreover, by discerning the areas where AI is already gaining traction, the AISE index may serve as a strategic tool for governments, labor organizations, and educators seeking to design interventions that enhance workforce resilience. Tailored training programs, adaptive regulatory policies, and forward-looking education curricula can be crafted around the nuanced understanding of AI adoption patterns unveiled by this research, ultimately smoothing the transition for workers facing changing occupational demands.
In summary, this study provides a vital recalibration of AI’s expected impact on employment by integrating entrepreneurial dynamics and social considerations into AI risk evaluation. It emphasizes that while AI is set to disrupt specific professions significantly, its influence will be asymmetrical, contextual, and mediated by human values and judgments. As AI technologies continue to evolve, policymakers and industry leaders must remain attentive not just to AI’s technical potential but also to the social frameworks that shape its real-world deployment and acceptance.
This nuanced understanding of AI’s potential impact on the workforce carries profound implications across the global economy. It urges a balanced, evidence-based approach that aligns technological innovation with human-centric values and societal expectations. By doing so, it offers a pathway for harnessing AI’s transformative capabilities while preserving the dignity and complexity of human labor in an increasingly automated world.
Ultimately, the insights procured through the Occupational AI Startup Exposure index underscore that the future of work in an AI-infused economy will be neither completely automated nor static but dynamically negotiated through the confluence of technology, society, and market realities. This study lays essential groundwork for anticipating and shaping that future, illuminating the critical factors that will govern the pace and nature of AI’s integration into the workforce.
Subject of Research: The study investigates the near-future exposure of various occupations to artificial intelligence technologies, emphasizing AI products developed by startups and their market viability as indicators of practical AI adoption across industries and job roles.
Article Title: Follow the money: A startup-based measure of AI exposure across occupations, industries, and regions
News Publication Date: 23-Jun-2026
Image Credits: Fenoaltea et al.
Keywords
Artificial intelligence, AI exposure, AI startups, labor market, occupational automation, large language models, LLM, Meta Llama3, venture capital, occupational tasks, AI adoption, AI socioeconomic impact.
