AI probably can’t do your job, a new study from Scale AI and the Center for AI Safety has found. Their research revealed that even the best artificial intelligence program could only autonomously complete 2.5 percent of projects it was given. The Remote Labor Index study asked a class of leading platforms, such as ChatGPT, Grok AI, Google’s Gemini, Anthropic’s Claude, and Manus 1.5, to complete a series of workplace tasks. They included creating floor plans, making games, and visualizing data. Programs failed half of the tasks by producing poor-quality work and were unable to complete a third of them entirely. Center for AI Safety director Dan Hendrycks said in a post on X, “While AIs are smart, they are not yet that useful: the current automation rate is less than 3 percent.” Researcher Jason Hausenloy told The Washington Post, “Current models are not close to being able to automate real jobs in the economy.” According to the Post, the research is a challenge to widespread predictions that AI will wipe out significant parts of the workforce. OpenAI boss Sam Altman said in July, “AI is for sure going to change a lot of jobs, totally take some jobs away, and create a bunch of new ones,” Yahoo Finance reports.
Read it at The Washington Post
