Why 80% of AI Projects Fail - and How to Make Yours Work

Commit & Push

Most AI projects fail. Some never ship. Others ship and implode. So what’s going wrong?
In this episode of Commit & Push, Damien sits down with Dan Saffer - Associate Director of Outreach at Carnegie Mellon’s Human-Computer Interaction Institute and author of Microinteractions - to dig into the real reasons so many AI projects fall apart.
Drawing on years of academic research and hands-on industry experience, Dan unpacks the five most common failure points: bad data, fragile models, vague value props, ethical landmines, and poor user adoption. They also dive into the myth of explainability, the broken state of AI UX, and why “sparkle-washing” products with AI features often makes things worse, not better.
Oh - and if you’ve noticed your favorite platforms slowly turning into garbage fire content mills? Dan’s got a name for that too: enshittification - and AI might be pouring gas on it.
Whether you're building AI tools, integrating them into your product, or just trying to separate signal from noise, this episode pulls back the curtain on what’s real, what’s hype, and what to do about it.
If you enjoyed this episode, make sure to subscribe, rate and review on Apple Podcasts, Spotify and YouTube, instructions on how to do this are here.
Episode Highlights:

• [00:00] Intro
• [02:00] Who Is Dan Saffer?
• [04:53] Why So Many AI Projects Fail
• [11:02] Why “Perfect” Use Cases Are a Trap
• [14:39] Trusting the AI Too Much
• [20:44] The Rise and Fall of Chat Interfaces
• [35:31] Enshittification: AI’s Role in Ruining Platforms
• [42:49] Stop Slapping Sparkles on Everything
• [44:52] AI Tools That Delight

Episode Resources:

• Dan Saffer’s LinkedIn
• CMU Human-Computer Interaction Institute Website
• Microinteractions Books
• Damien Filiatrault on LinkedIn
• Scalable Path Website