The researchers applied machine learning models to measure the semantic similarity between task descriptions and patent descriptions. This analysis went beyond simple word matching, comparing entire task descriptions to entire patents. For each task, the most similar AI patent was identified, and if the similarity exceeded 90%, that task was considered impacted by AI. The researchers assigned each occupation an AI Impact (AII) score, based on the number of tasks influenced by AI patents divided by the total number of tasks for that occupation.
The study highlighted both the most and least affected jobs by AI. Occupations such as orthodontists, security guards, and air traffic controllers were among those most likely to be impacted, while roles like pile driver operators, dredge operators, and agricultural product graders were least affected.
Interestingly, the research revealed that jobs involving repetitive tasks were not always the ones most impacted by AI. Instead, tasks requiring a specific sequence of actions to produce machine-readable outputs were more susceptible. For instance, AI's potential role in analyzing medical images, such as X-rays, could assist doctors in identifying health issues but wouldn't entirely replace healthcare professionals.
The team concluded that fears of AI eliminating large numbers of jobs might be overstated. Many sectors, such as healthcare and transportation, are facing worker shortages and may benefit from AI augmenting human labor rather than replacing it. They stressed that AI developers should focus on creating tools that enhance human capabilities, and industry leaders should emphasize education and training to ensure that workers can adapt to and benefit from AI technologies.
Research Report:The potential impact of AI innovations on US occupations
Related Links
Nokia Bell Labs Cambridge
All about the robots on Earth and beyond!
Subscribe Free To Our Daily Newsletters |
Subscribe Free To Our Daily Newsletters |