This echoes the MIT study that shook markets in August with the finding that 95% of generative AI pilots were failing.