Why 85% of AI Projects Fail (And How UX Can Prevent these Costly Mistakes)
How storyboarding can help ensure that AI experiences serve user needs
Jakob Nielsen, in his prophetic vision of the AI-integated design future, was partially correct.
The real question around AI isn’t what it can do. It’s what users actually need it to do.
This is why, according to Gartner, 85% of AI projects fail, and why designers who understand AI have a bright future. Why?
According to Greg Nudelman, author of “UX for AI,” the primary culprit is simple: the use case is bad.
Why Use Cases Matter More Than Technology
Nudelman uses the example of “boiling spaghetti”, a close analogy to a technical system used to purify natural gas.
Anyone who’s cooked pasta knows this can be tricky. Pots suddenly boil over, spilling water and creating a mess everywhere. In Nudelman’s case, gas would “boil over,” costing the company $100,000+ in cleanup.
Experienced operators handle this intuitively through trial and error. Then the company tried to replace them with AI: an AI use case bound to fail. Why?
The disconnect often lies between business and user use cases. The business case seemed valid: expert monitoring costs $500,000 annually, creating pressure to automate. But if the AI doesn’t solve a real problem for the people who need to use it, the project fails.
The AI Art Disconnect
Nowhere is this clearer than with AI art generation. From a business perspective, AI art appears highly effective.
Companies don’t need to:
Hire dedicated artists
Maintain long-term creative staff for consistent imagery
Manage contractor relationships
Invest in creative talent acquisition
Handle complex creative project management
But why would users actually want AI art?
Sure, you can experiment with creating Marvel characters in Disney style, but what underlying problem drives users to seek this solution? Without a real problem, AI becomes merely a novelty.
This “lack of usage” translates into real business harm: loss of user trust, low adoption and retention, poor ROI. This is why designers who can translate user perspective into strategic value are critical.
Storyboarding Reveals the Truth
Showcasing the lack of a use case is easier than you think. All you need is a storyboard.
Look at Nudelman’s sketch: an unhappy person at a coffee shop opens an app and becomes happy. Can you tell what the product is?
The sketch is about a Mental Health app, but that’s not immediately apparent. The app could be anything: social media, stock trading, or a to-do list.
This is the first immediate problem a storyboard solves: being specific with the actions users take. When mapping out the journey, do the steps clearly highlight the product use case from start to finish?
Testing the Use Case
Once you have your storyboard, it’s time to stress-test it.
Here’s Nudelman’s explanation: “A young man sitting alone in a coffee shop feels deeply depressed, so he launches the Mental Health Assistant, an AI-based app with questions a licensed therapist might ask. Through answering these questions, he feels much better.”
Ask yourself:
Would you seek mental help in a public coffee shop?
Would answering a list of questions really make you feel better?
Could you treat depression that easily with an app?
Would users open your app instead of binging on social media?
These questions reveal why this is a doomed use case. It doesn’t matter how excellent the app is if users don’t behave the way the business expects.
Redirect Toward a Better Use Case
As a designer, you probably can’t cancel a project based on a flawed use case. But you can offer a better alternative that matches user needs.
Here’s how Nudelman redirected the app: “A young man sitting alone in a coffee shop feels shy but wants to approach a woman he’s been wanting to meet. He launches the Mental Health Assistant, which encourages him to be less afraid of rejection. The woman is happy to talk, they strike up a conversation, and romance blossoms.”
What changed:
Tackling mild social anxiety instead of depression—something an app can actually help with
Users who rely on external motivation (self-help books, YouTube videos) already exist
A coffee shop is an ideal meeting place where pulling out a phone isn’t strange
This use case doesn’t just meet business needs: it’s something users might actually do.
How to Run an Effective Storyboarding Session
The beauty of storyboarding is its simplicity, but running a practical session requires some structure. Here’s how to set yourself up for success:
Get the right people in the room. You need someone who understands the technical capabilities (a product manager or tech lead), someone who knows the business goals (a stakeholder), and ideally someone who’s talked to users recently. Keep it to 3-5 people maximum—too many voices and you’ll never finish.
Start with the user’s context, not your solution. Ask: Where is this person when they encounter this problem? What just happened before? What are they trying to accomplish? This grounds everyone in the user’s reality before jumping to your AI solution.
Sketch the journey panel by panel. Use post-its or a whiteboard. Each panel should show an explicit action or transition. Don’t worry about artistic quality—stick figures work perfectly. The goal is to make the user journey visible and concrete.
Pause at each transition and ask “why?” This is where you’ll catch flawed assumptions. “Why would they open our app instead of YouTube?” “Why would they do this in public?” “Why would this make them feel better?” If the answer is “because our product is great,” you’ve found a weak point.
Reframe, don’t reject. When you spot a flawed use case, resist the urge to say “this won’t work.” Instead, ask: “What would need to be true for a user to behave this way?” or “Is there a similar context where this makes more sense?” This keeps stakeholders engaged rather than defensive.
The whole exercise should take 30-60 minutes. Any longer and you’re overthinking it. The goal isn’t perfection. It’s exposing assumptions before they become expensive mistakes.
The Designer’s Role in the AI Era
Despite the hype around AI tools, the future of UX remains unchanged: considering user needs. This matters especially when businesses pursue sophisticated features while neglecting their users.
Strategic designers are needed more than ever. We’re the advocates for user needs in a landscape where technical ambition often overshadows practical utility.
The tools to do that? Sometimes they’re not high-tech. Sometimes it’s a simple sketch on a post-it that spells out the user journey from start to finish, ensuring that AI development serves human needs rather than just making businesses appear innovative.
Kai Wong is a Senior Product Designer and Data and Design newsletter author. He teaches a course, Data Informed Design: How to Show The Strategic Impact of Design Work, which helps designers communicate their value and get buy-in for ideas.