HR Leaders Why 90% of AI Projects Fail (and How to Fix It)
Nov 14, 2025
Eric Mosley, Founder and CEO of Workhuman, dives into the reasons why 80-90% of AI projects fail, attributing it to a lack of meaningful data. He highlights how recognition data can fuel AI innovation by uncovering hidden talent and reducing bias. Eric shares insights on the emotional ROI of gratitude at work, the importance of mobile and peer-to-peer recognition, and the potential for AI to predict career paths. With a vision for a future where AI amplifies humanity, he reveals actionable steps for building effective recognition programs.
AI Snips
Chapters
Transcript
Episode notes
Data, Not Tech, Drives AI Success
- Most AI projects fail because companies lack meaningful training data, not because of the technology itself.
- Recognition data captures real performance, skills, contribution, and relationships that boost AI accuracy.
Recognition Data Is Gold For AI
- Recognition programs produce rich natural-language signals about employees that NLP and LLMs can decode into skills and performance signals.
- That data becomes actionable intelligence to find talent and measure contribution across an organization.
Treat Recognition As Strategic Data
- Build a data science capability and analyze recognition moments to surface insights rather than treating recognition as a checkbox.
- Embed those AI models into the recognition flow so managers get instant, context-rich answers when they need them.