The next phase of AI must start solving everyday problems
As Anthropic and OpenAI duke it out with Pentagon matters, Cowork capabilities, and model launches, it’s important to remember that technology is not the goal. It is a means to an end. Its value comes from helping people solve daily problems and giving them one less thing to think about—on a global scale.
However, people must first realize there’s a problem and understand how technology can solve it before AI can make a meaningful difference.
When things click, it’s always a matter of consumer education, which leads to expanded adoption, which in turn leads to society-wide impact (in that order). Each step can happen swiftly—or take months or years to complete.
This pattern—education first, adoption next, transformation last—repeats across sectors. It’s also a tale as old as (human) time.
The lesson from past cycles like the cloud and mobile web: The best AI-powered systems won’t be the ones with the highest investment totals or most bells and whistles; they’ll be the ones with tech that unceremoniously makes real-world processes faster, cleaner, cheaper, and more resilient.
Technology adoption at scale isn’t an overnight phenomenon; it’s a signal that technology has crossed a threshold from curiosity about “the new thing” to daily driver. AI has tremendous potential, but we in the tech world have lost the plot on making this matter for real people. I’ve been fortunate enough to see this cycle play out a few times in my career.
For example, when I worked on the first iPhone, it was impossible to predict a future powered by dating apps, ride-sharing, mobile payments, or social media. Now, it’s hard to remember a time before we could run our lives through our phones. Our breakthrough was delivering an ecosystem once the tech was powerful enough and the world was ready. Because of the backbone we created, new platforms emerged that allow people to leave their wallets at home and conveniently pay via their phones, or tap a button to get a ride.
Once consumers realized the power and ease of solving real-world problems with a swipe or tap of their fingertips, adoption took off like wildfire.
The same principle was true when we built the first Nest thermostat. From the beginning, the goal was to apply technology to make energy more efficient, from both a capacity and cost standpoint, for households and regular people. We talked about building AI-powered devices that could understand human behavior and adapt accordingly. We had the vision, but needed AI to advance in order to make progress technologically possible and developmentally practical.
For example, a popular, seemingly humble feature like package detection on a Nest doorbell camera took nearly a year to develop. The models were heavy. The hardware was constrained. The development cycles were long. We were crawling toward a clear goal with hundreds of hurdles in our way.
By the time we perfected computer vision over the course of more than a year, consumers understood the problem that Nest was solving and how adopting the system would help them reduce utility costs while using energy more efficiently. It was at this point that the transformation at scale could—and did—happen.
But it takes more to scale meaningfully than shipping innovation and pushing updates to consumer devices. You need to combine the latest technology with the level of consumer interest to solve the problems we face on a daily basis.
At Mill—the food recycler company I now run—we started by focusing on households, helping people manage food scraps and deliver them back into the food system, one kitchen at a time. That phase mattered. Education that leads to behavior change always comes first. People must realize there’s a system-wide problem and understand why it exists before tech can help solve it.
Food waste, for example, is an industrial problem. Grocery stores discard millions of pounds of food every day. Behind every supermarket is a loading dock crowded with dumpsters and compactors that are consuming space, energy, and labor—and these valuable resources end up getting sent to a purgatory of methane production.
Developing an enterprise-grade, AI-powered food waste system at Mill, and seeing it adopted by major players like Amazon and Whole Foods Market, is a signal that we’ve entered a new phase. It’s clear that reducing food waste isn’t just about nudging individual habits. It’s not just about putting last night’s lasagna in the right bin. It’s about removing entire classes of waste from the system.
Artificial intelligence makes that possible—not because it’s flashy, but because it’s finally reliable, affordable, and fast enough to operate at scale in the physical world.
With the iPhone, smart devices like Nest, and now AI, perspective matters. But above all else, tech leaders need to keep in mind that we must be solving real problems—not generating tech for tech’s sake.
Progress in this physical era of AI requires logic and restraint as much as ambition. Fortunately, we’ve been here before. The internet was speculated to become a Wild West of lawless virtual worlds and digital avatars. It became a functional tenant of digital society, grounded by email, maps, commerce, and communication—mundane tools that solved ordinary problems at an unprecedented scale.
The story of AI’s next chapter is steeped in precedent. Hype will fade. Models will commoditize. Launches will grow quieter. What we’ll hopefully be left with are AI-powered systems that work to solve everyday problems while improving life in the physical world, rather than distracting from it.