Benedict Evans, a long-time tech analyst, isn't here for the AI hyperbole. He's also not here for the dismissiveness. His "most controversial opinion" is a tightrope walk: “I think that AI is as big a deal as the internet or mobile and only as big a deal as the internet or mobile.” For founders in their 20s and 30s, who’ve lived their entire lives post-internet, it's a critical reframing. We tend to inflate the new and diminish the old, forgetting the sheer scale of past shifts.
We’re Living In 1997 Again
To understand AI today, Evans urges us to think back to 1997. It wasn't that the internet wasn't exciting then; it was just incredibly messy. “Most stuff kind of doesn't work yet,” he says, reflecting on that era. Similarly, most truly impactful AI applications haven't been built yet. The core technology, like large language models, feels cool, but widespread, indispensable use cases are still hazy. “Most people are using who are using this are using this every week or two maybe,” Evans notes about current AI adoption, highlighting just how early we are.
Think about 1997: Google wasn't founded until 1998, Facebook until 2004, and the iPhone wasn't released until 2007. The internet was a novelty for many, a tool for early adopters, and its full economic and social impact was still decades away. The same applies to AI. The current crop of AI tools are like the clunky websites and dial-up modems of '97 – hinting at a future but far from the seamless experiences we expect today.
The True Scale Is Easy To Miss
It’s tempting to look at AI and demand it be more than the internet or mobile. But Evans reminds us of our collective amnesia for truly giant shifts. “We forget how big a deal the internet was,” he explains. “We've had these absolutely enormous changes and then we don't see it because it's like that's the world the world has always been.” The internet completely rewrote how we communicate, shop, learn, and govern. Mobile put that entire revolution in our pockets, changing physical spaces and human behavior. AI, by allowing machines to understand, reason, and create, is a change of similar gravity, not necessarily an order of magnitude beyond it. It will reshape industries, create entirely new ones, and profoundly alter daily life, but like its predecessors, its most radical effects will only become apparent once they're so ingrained they're invisible.
This isn't to say AI is less exciting, but rather that its impact will be absorbed into the fabric of daily life over years, not months. The biggest wins won't come from chasing speculative, sci-fi visions, but from solving real problems within this '1997' context. As Evans puts it, a presentation on AI often boils down to: “This is 80 slides are saying we don't know which is like slightly facitious but also kind of true.” The uncertainty isn't a bug; it's the feature of a genuinely new platform.
What to Do With This
Stop chasing the bleeding edge of foundational models. Instead, act like it's 1997 and focus on building practical, user-facing applications that solve clear problems, even if they feel clunky or niche at first. Recognize that widespread adoption will take time, so build for genuine utility and iterate quickly, rather than waiting for a perfectly mature ecosystem that doesn't exist yet.