A new study uses machine learning to reveal which jobs are most and least impacted by AI innovations, providing crucial insights for workforce planning and tech development.
Artificial intelligence (AI) has long been heralded as a game-changer across multiple industries, yet its precise influence on individual jobs remains a contentious topic. A new study promises to shed light on this issue by using AI to evaluate its own potential impact on various occupations.
The research, conducted by Daniele Quercia and colleagues from Nokia Bell Labs, employed machine learning to analyze a vast dataset of 17,879 occupational tasks listed in O*NET, a U.S. government-run occupations database.
This dataset was cross-referenced with 24,758 AI patents filed with the United States Patent and Trademark Office between 2015 and 2022. By measuring the semantic similarity between task descriptions and AI patent descriptions, the study created an AI Impact (AII) score for each occupation.
The team’s analysis did not rely just on word matching but actually compared entire task descriptions to entire patents, as noted in this news release.
The AII score was determined by identifying the most similar AI patent for each task and evaluating if this similarity surpassed the 90% threshold. If a task met this criterion, it was deemed impacted by AI. The researchers then calculated an AII score for each occupation by dividing the number of tasks influenced by AI by the total number of tasks in that occupation.
The study revealed that some of the most impacted occupations include orthodontists, security guards and air traffic controllers. Conversely, occupations such as pile driver operators, dredge operators and graders of agricultural products were found to be least impacted by AI technologies.
Contrary to some predictions, the study found that repetitive tasks were not always the most affected. Instead, jobs that involve a specific sequence of tasks producing a machine-readable output were more likely to be impacted. This insight dispels some of the more alarmist forecasts about AI’s role in the workforce.
The study also found that “research often overestimates how much AI will take away jobs because sectors potentially impacted by AI, like healthcare and transportation, currently need more workers, not fewer and because AI is likely to augment rather than replace many jobs,” as noted in the news release.
For example, while AI can analyze medical images such as X-rays to assist doctors in identifying health issues, it cannot replace the nuanced expertise and decision-making capabilities of medical professionals. This highlights the potential for AI to augment rather than replace human labor, particularly in fields requiring complex judgment and human interaction.
The authors emphasized that tech developers should prioritize creating AI tools that support people rather than replace them and called for leaders to focus on training and education to maximize the benefits of AI.
The full study, published in PNAS Nexus, offers invaluable insights for policymakers, tech developers and the workforce at large, providing a roadmap for integrating AI into various sectors without displacing workers.
For a detailed visualization of the study’s data, check here.
In an era where the role of AI is rapidly evolving, this study provides much-needed clarity, offering a balanced perspective on the future of work in an AI-enhanced world.