When the AI gold rush ends

Summary

We’re living in an AI gold rush, where product teams are pushing at the fringes of tech for fear of missing out. But in the process, are we ignoring the core of our products and the majority of users?

Most product managers have, at some point, read about the innovator’s dilemma. Do you improve your existing product or push for value at the unproven fringes? If you’ve read the late great Clayton Christensen’s books, you probably know all the examples he quotes of companies that faced this dilemma and made the wrong choices. Kodak, Blockbuster, Nokia… there are many such horror stories.

My biggest takeaway from Christensen’s books was that there’s always unproven technology on the fringes, which someone can exploit to disrupt an incumbent someday. Indeed, many product managers run a two-track development process to guard against such disruption. One track handles the business as usual, core delivery, while the other runs experiments to prove new ideas on the fringes. Ideas are cheap, execution is costly. The “discovery track” helps us learn, at low cost, if an idea is even worth pursuing. 

But while the ideas on the fringe can often differentiate a product from its competitors, most users derive value from the core features of a product. In the last two years, I’ve noticed that the AI revolution has spotlighted this product development paradox. The tech has improved at a frightening pace, with every iteration of the frontier models. As that’s happened, product teams have found themselves with more ideas than they can execute. There’s FOMO, the fear of disruption, and the dread of becoming obsolete or irrelevant. 

AI needs training data. The bigger the company, the more likely it is to have access to such data. Unsurprisingly, many of the popular LLMs have the backing of either a hyperscaler or the likes of Elon Musk or Adobe. Smaller companies face an existential, “innovate or die” threat, in the face of how these behemoths are now expanding their product suites using AI.

So today, more than ever, you see product teams rushing to add more AI features and generative capabilities to their products while neglecting the core of their product, where most users derive value. 

Product value lies at the core, but teams chase it at the fringes

OpenAI unleashed the AI revolution when it released ChatGPT in November 2022. AI is an indispensable part of our workflows today. Yes, the world is talking about agentic workflows, multi-agent ecosystems, self-directed autonomy and artificial general intelligence, but remember that the average user still struggles with prompting! And several products do little to improve this human-machine interface and chase value at the fringes.

In an era where every other product sits on a dot ai domain, the FOMO presents an interesting product management challenge. The frontier models will keep getting better, and the researchers of these models will continue to tell us how these models will solve new, more complex problems. No doubt, they will. Most users, though, may struggle to adopt such new technology, as they do today. 

Investing in the needs of the majority

Someday, in the near future, the AI dust will settle. Innovation will continue, but AI will be table stakes, much like mobile apps, cloud, and UX were at one point. And when the gold rush ends, I hope we product managers can take a breath and improve the core of our products instead of constantly pushing at the fringes. Maybe once again, we can focus on the early and late majority of users, not just on the innovators and early adopters. I’d certainly welcome a saner, calmer, slower, more thoughtful world of product management, where making our users awesome will again be more important than making the product awesome.