AI Job Disruption Will Be More Gradual Than Feared, MIT Study Finds
-
New MIT study finds that majority of jobs previously identified as at risk of AI automation are not currently economically beneficial to automate.
-
Researchers found only 23% of wages for visual tasks would be attractive to automate with AI - humans still more cost effective.
-
Even with rapid cost decreases, could take decades for AI automation of jobs to unfold - time for policies.
-
Study limitations doesn't consider AI augmentation of jobs, job creation, or all cost savings from pre-trained models.
-
Key takeaway is that AI job disruption may happen slower and less dramatically than suggested.