The study also highlighted that even with self-hosted, self-service AI systems that only need fine-tuning, many jobs, particularly low-wage and multitasking-dependent ones, wouldn't make economic sense for a business to automate. The researchers acknowledged the study's limitations, including not considering cases where AI can augment human labor or create new tasks and jobs. They concluded that while it's important to prepare for AI job automation, the process will take years or even decades to unfold.
Key takeaways:
- A new research study from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) explores the potential of AI to automate human jobs and the economic feasibility of such automation.
- The study found that the majority of jobs previously identified as being at risk of AI displacement aren’t, in fact, “economically beneficial” to automate — at least at present.
- The study only looked at jobs requiring visual analysis and did not investigate the potential impact of text- and image-generating models on workers and the economy.
- Despite the potential for AI to automate tasks, the researchers suggest that the coming AI disruption might happen slower — and less dramatically — than some commentators are suggesting, and there is time for policy initiatives to be put into place.