Can we predict the jobs robots will do? (www.anthropic.com)

A new study introduces a robot exposure index that evaluates how well robots can perform various job tasks, revealing that current robotics technology can handle about 74% of physical tasks in the U.S., accounting for 34% of working hours. This capability predominantly affects jobs like driving and warehousing, where robots are already being utilized, while fields like nursing remain largely unaffected due to the complexities and interpersonal skills required. Notably, even though robotics has made significant advances, they are only cost-competitive for approximately 0.3% of job tasks, suggesting that it could take up to 40 years for this share to rise to 10% as costs decline. The implications of this research are profound for the AI/ML community and the workforce at large. It highlights the importance of understanding current robot capabilities as a means to predict which jobs will be at risk in the future. Historical analysis indicates that occupations with higher exposure to robots have previously faced declines in wages and employment. Therefore, as robotics technology continues to evolve, jobs that robots can already perform in controlled environments are likely to see the earliest disruptions. By establishing a method to gauge job exposure to existing robotic capabilities, this study lays the groundwork for anticipating future automation trends and preparing for their socioeconomic impact.