MIT Study – 95% of Generative AI Investments Fail (or do they?)

MIT released a study showing that 95% of organizations are getting zero return from their GenAI investments.

While some may claim this proves AI is all hype, a closer reading suggests the findings aren’t a death knell for the technology. Instead, they reveal critical truths about how to succeed.

🎯 It’s Goodhart’s Law writ large
The report shows GenAI adoption is driven by areas like sales and marketing, where success is easier to measure. Pilots are optimized for visible, top-line metrics. However, the study suggests the most dramatic cost savings come from the back office – reducing BPO contracts and agency spend, where the ROI is clear but less flashy.

💡Knowledge and memory are sensitive to each organization
General-purpose AI tools will never work perfectly because each company has its own ontology, its own ways of making sense. Building tools sensitive to this is critical. But there is a contradiction – the study finds that these highly-contextual internal projects fail twice as often as those led by external partners. This is the gap where a strategic partner can help bridge deep internal context with external expertise. (🙋🏾‍♂️ – Jeavio)

🤨 There is a productivity paradox at the heart of GenAI adoption
Workers from over 90% of companies surveyed reported regular use of personal AI tools. If individuals are seeing productivity gains, why does it fail at the aggregate? The report suggests the reason is simple: the most successful AI adoption is bottom-up, not top-down. Successful organizations source initiatives from “frontline managers” and “power users,” not central labs.

At Jeavio, we live this principle. We host hackathons and sponsor open-ended projects to explore how AI can address real-world problems. The ADAPT platform, our flagship AI initiative, began as an internship project in 2023.