Why Most AI Projects Fail: Insights from UX Design Expert Saloni Pasad

Saloni Pasad is a Senior UX Designer with an M.S. in Human-Centered Design and a Bachelor’s in Communication Design. She works with global clients whose platforms serve millions of users, where the stakes are high and her creative design choices impact not just user experiences, but also the success of entire businesses.

Despite billions in investment, many AI projects flop, and it’s rarely because the technology is bad. We spoke with Saloni to uncover what’s often missing in AI design.

1. 
   Why do so many AI projects fail, even when the technology is advanced?
   
   A. Artificial intelligence is often hailed as a magical tool that can transform our work, our products, and even our lives. But in reality, many AI projects fail, not because the technology is bad, but because the creative thinking around it is missing the mark.

From a designer’s perspective, the problem is rarely technical. It’s about human value: AI is most powerful when it solves problems that truly matter to people, yet too often, projects chase flashy features or overhyped ideas that feel innovative internally but don’t resonate with real users.

1. What’s missing when AI promises innovation but doesn’t deliver?
2. At the heart of AI’s creative struggle is what I call the innovation gap. Designers think in terms of desirability: what will delight users and make their lives better. Data scientists often think in terms of feasibility: what the technology can do. The sweet spot, where AI is both useful and possible, is where these two perspectives intersect.

I’ve seen teams pour months into creating clever AI features that are technically brilliant but don’t solve meaningful problems. Inside the lab, it feels innovative. Outside, it falls flat. The result is frustration, wasted effort, and lost opportunity.

1. Even when it’s not perfect, where can AI really make a difference?
2. AI doesn’t need to be flawless to be valuable. Some of its most elegant applications enhance human creativity rather than replace it. For example, Gmail’s Smart Compose helps you draft emails faster, Spotify’s Discover Weekly suggests music you might love, Google Photos curates vacation highlights resurfacing forgotten gems, and Apple Photos lets you search for a ‘dog at beach’ to instantly find that one photo.

These tools don’t try to do everything perfectly, they suggest, inspire, and accelerate, leaving humans in charge. Think of AI as a novice collaborator: eager but clumsy, inconsistent but surprisingly helpful in the right contexts. You wouldn’t let a beginner drive a plane, but you might let them mix music at a party. Similarly, AI shines in low-risk, repetitive, or time-consuming tasks, freeing human creativity for higher-order work.

1. From your experience, what patterns cause AI projects to stumble?
2. Through a design lens, AI failures often follow predictable patterns:

1. Some AI features fail because they just don’t solve real problems for users.
2. Tools that don’t generate revenue, improve efficiency, or support business goals are quickly deprioritized or abandoned.
3. When companies rely on scraped social media data or biased datasets to train their AI.
4. When the AI model fails in real world conditions and can’t make accurate predictions. 
5. When AI doesn’t feel fair, transparent, or human-centric, it erodes trust.

1. How can we design AI that actually enhances creativity and human value?
2. The most compelling AI projects don’t chase technological breakthroughs for their own sake, they focus on human-centered problems that AI can meaningfully augment. The criteria are simple:

1. The AI solution should be of High user value. Does it make life easier, more enjoyable, or more creative?
2. It should be low risk. Can people still succeed if the AI makes a mistake?
3. Its technical demands should be moderate. Is the solution feasible without forcing perfection?

A tool that helps a designer sort assets or suggest creative ideas doesn’t need to be perfect. It just needs to be better than the default, helping humans amplify their own abilities.

1. Looking ahead, what’s the future of AI when we put humans at the center?
2. The next wave of AI innovation will come not from pushing models to their technical limits but from rethinking how we design for human value. It’s about choosing the right problems, embracing imperfection, and creating AI tools that collaborate with humans instead of trying to replace them.

AI won’t replace people, but misaligned, poorly designed AI projects can stifle creativity, waste time, and even threaten the organizations that build them. The key is creative partnership, not technical bravado. When AI and design meet in that sweet spot, the results can be transformative, unexpected, and profoundly human.