A study came out this week stating that 95% of all AI initiatives lose money or fail to achieve their goals. Specifically, the study was referring to organizations building AI-enabled products, services and systems. While not a quantitative study over a large sample size the study was big enough to identify a trend — as we’ve done time and again, companies are throwing good money after bad when it comes to AI. This is good news for product managers and designers. 

Gen AI makes progress look easy

I have to admit that I’ve been super impressed with some of the output I’ve seen from a variety of Gen AI tools that proclaim to do all the tasks designers and product managers normally do in a fraction of the time. At surface level, the PRD templates (heck, completed PRDs), wireframes, prototypes, and even working code give the impression that the work is done. And not just done, but done in minutes. That’s where the problems begin. 

Context, intuition and taste can’t be faked

The Gen AI tools we have at our disposal today respond to natural language prompting as well as text-based artifacts to provide answers that mimic what “good” should look like. What these tools can’t mimic is a deep understanding of the context of the work it’s producing. What drove the initiative in the first place? What problem are we trying to solve? What have we tried in the past? What patterns have we seen in the market? The AI doesn’t have any of that. Could you feed it to it? Sure, but again, the output will yield lowest common denominator solutions. 

How many times have you gone with your gut on a product decision? My guess is more than a few. Gen AI tools have never done that. They don’t have that gut feeling, that intuition that one direction is worth exploring over another. Could Gen AI have given us the iPod and music ecosystem that was built around it? How would it have known to pursue this particular pain point in human interaction? 

Finally, there’s taste. Pattern matching yields homogeneous results. I recently ran the same Gen AI conversation with Claude and ChatGPT. The results overlapped by more than 80%. Best practices are a great place to start but they don’t account for taste. Innovation comes from taste and, for now, you can’t teach that to a bit of code. 

Product management and design — now,  more than ever

With 95% of AI initiatives failing, this is the moment for product management and design to shine. The successful initiatives we launch in the coming years, built on the back of these new AI capabilities, will be the result of experience, expertise, intuition and taste. They’ll embody the human element that algorithms can’t emulate. More than ever, we need to make sure there’s a meaningful problem to solve and that AI makes sense as a part of the solution. We need to push beyond the bland outputs that will get us off the blank page faster than ever but won’t get us to a winning solution. And we need to make sure that the results our AI-powered services make sense for our users. It’s a new spin on an old formula. The tools are faster and more exciting than ever. The humans we serve, however, still need us to make them successful.