Keeping It Human in a Digital World (6)

First, we simply haven’t had enough time to fully absorb its capabilities. Although billions have access to sophisticated AI, only a small fraction have meaningfully integrated it into their daily lives. But we do know that interest in experimenting with AI is already surpassing the early days of other transformative technologies. A recent study by the Federal Reserve Bank of St. Louis, in collaboration with Vanderbilt University and Harvard Kennedy School, found that nearly forty percent of people aged eighteen to sixty-four have used AI. To put that in perspective, it took the internet two years to reach just a twenty percent adoption rate and three years for the PC to do the same.

These signs of early adoption suggest that, as more people discover how to get real value from the technology, its widespread use could take off rapidly. Another crucial factor is that entrepreneurs are only just beginning to leverage AI in the products and services we rely on for work and personal use. Though there are already a staggering number of AI-powered products available, we are just at the beginning of this innovation cycle, with much more to come our way. One day, many of us will rely on AI features just as much as we currently rely on smartphone apps (which went from being new to indispensable in just over a decade).

What about from the perspective of businesses that are not in the tech industry? Here too we are only in the earliest stage of the AI era. Companies are just beginning to develop the talent and processes needed to use AI effectively. They are just starting to learn how to set up the organizational structures to support innovation in and responsible use of AI. Only a handful of external-facing applications are in meaningful production or at any significant scale. While AI models have already ingested vast public datasets, most organizations have barely begun to tap into their proprietary data. As companies strengthen their AI capabilities, they’ll be able to make better use of today’s technology and become more adept at developing their own AI products and services. This could lead to entirely new categories of solutions that are hard to foresee from where we stand now. And once they demonstrate real paths to success, we’re likely to witness many more followers—a gold rush of companies racing to capitalize on these insights and secure their place in an AI-driven future.

But another factor is at play: over the last few years, we’ve experienced breakneck innovation in the underlying technology. Researchers found that Large Language Models (LLMs) had been improving several times faster than Moore’s Law, a critical engine that drives tech acceleration. Moore’s Law predicted that we would double the capacity of a computer chip every two years. While this pace has slowed, it explains the miraculous reality that your smartphone—probably within arm’s reach right now—holds more computing power than the massive computers that once filled entire rooms. The 2023 study showed that Large Language Models (LLMs) were improving faster than Moore’s Law: the computing power needed to advance AI was halving every five to fourteen months. That’s a speed of advancement that stunned even seasoned technologists.

You can easily see the difference in AI sophistication from a year or two ago by comparing the results of the same prompt over time. Meanwhile, researchers track AI’s progress more precisely through rigorous performance tests. While there are compelling arguments that this evolution will plateau (or already has), we’re seeing pioneers push different levers to keep advancing the space. Many of the world’s most powerful companies are continuing to make eye-popping investments in this fiercely competitive race, constantly finding new ways to propel the technology forward and jockey for market lead. New and different kinds of models are emerging, including smaller models that work better on devices right where they’re needed (such as your phone) and models optimized for specific industries or domains. Adding to this momentum is another critical factor: artificial intelligence is getting to the point where AI is developing AI. In other words, it is starting to build and improve itself, which could propel further advances.

Why Now is the Best Time to Begin Your AI Journey

What does this mean to you? Regardless of the exact rate of improvement, it’s a rapid pace of development—the fastest I’ve seen over a career in Silicon Valley—and AI is continuing to get better at what it does. A common refrain bounces around the AI community: the AI you use today is the worst AI you will ever use. And this will be true next month, next quarter, and next year too.

This means that people and organizations who already know how to use AI effectively will continuously increase their advantage. Catching up is only going to get harder.

But today, AI isn’t yet delivered in a way that really works for the average person. What looks like a finished product—for example, a chatbot such as ChatGPT—lacks the user-friendly guidance and refinement we expect from typical consumer products. There are some notable exceptions—companies that have packaged AI into products that add significant value to a specific workflow—but this is still quite new. As technology analyst Benedict Evans explains, “A chatbot looks like a product. You type something in and you get magic back! But the magic might not be useful, in that form, and it might be wrong. It looks like product, but it isn’t.”

This gap is precisely why we need this book. We’re so early in AI that its creators have essentially left us to figure out how to use it effectively on our own. But this demands a fundamental shift in how we interact with software—a change that’s far from intuitive for most people.

Part Three of this book guides you on how to engage with AI as early adopters do. These pioneers aren’t waiting for polished AI products—they’re jumping in now and forging their own path to value. As you embark on this journey, think like an early adopter: don’t judge AI by its current limitations, but learn to work around its quirks and anticipate “where the puck is headed.” The time to start is now, even in this raw, early stage. Those who delay may find themselves facing an increasingly unbridgeable gap as AI expertise compounds and the technology races forward.

Notes