The Internet's Skeptics Were Wrong About Adoption, and Right About Governance

The internet's rise ran through a sequence that should feel familiar by now: early skepticism, uncontrolled experimentation, infrastructure strain, moral panic, and eventual absorption into nearly every corner of life. We have seen the same movie a few times since, most recently with AI.

The numbers show how far it traveled. In 1995, Pew found that only 14 percent of US adults had internet access, and plenty of Americans had either never heard of the internet or had only the haziest sense of it. By early 2000, about half of US adults were online. Today 96 percent of US adults say they use it.

From near-novelty to near-universal in a single generation. Hold that trajectory in mind, because the people who were loudest at the start were often the most confidently wrong about where it would end up.

The skepticism was not lazy. In a 1995 Newsweek essay, Clifford Stoll called the internet an ocean of unedited data and a wasteland of unfiltered data, and argued that users would struggle to tell what was worth reading. Robert Metcalfe, the Ethernet pioneer, predicted the internet would go spectacularly supernova and catastrophically collapse in 1996.

When the collapse failed to arrive, Metcalfe famously, and literally, ate his words. The adoption predictions were just wrong. The internet did not fail, did not implode, and did not stay a niche hobby for the technically inclined.

It is tempting to file these away as cautionary tales about doubting new technology. That is the wrong lesson to draw, and drawing it leads organizations to wave off legitimate concerns about AI by pointing at the people who underestimated the internet.

Here is the part that tends to get dropped: those predictions were wrong about adoption but far from useless. They named real transition risks, including information overload, broken trust, capacity limits, and shaky business models. The internet did not collapse, but it did give us spam, fraud, misinformation, cybercrime, privacy loss, platform dependence, and a steady supply of new manipulations.

So the skeptics were wrong that the technology would fail and frequently right that uncontrolled adoption would cause serious institutional problems. Both are true at the same time, and holding them apart is the move that makes the rest of this clear.

Stoll mistook a governance problem for an adoption problem. His ocean of unedited data was real enough. It just was not a reason the internet would stall; it was a preview of something we would spend the next thirty years building institutions to manage.

This distinction is essential for AI. Doubting that AI will be adopted and doubting that AI can be governed well are two different positions. Someone can be flat wrong that AI will not catch on and still be right that it will create reliability, bias, security, privacy, labor, intellectual-property, and accountability problems.

Organizations should not brush off governance concerns just because earlier skeptics lowballed adoption. Nor should they block adoption because risks exist. Both moves jam two separate questions into one and land on the wrong answer.

The disciplined approach is to ask them one at a time. Will this technology spread? Almost certainly. What controls does safe, productive use require? That is a real and answerable question, and it deserves its own serious work instead of being settled by the answer to the first one.

The internet also shows how crude the early business models tend to be. In the 1990s, most organizations treated a website as a brochure. Retailers dabbled in e-commerce long before logistics, payments, personalization, and customer service were ready for it. Media companies put their content online without much idea of what it would do to their revenue.

Enterprises wired systems together before they understood their own cybersecurity exposure. The technology spread faster than the governance, the operating models, and the economics around it. Capability outran the institutions, and the institutions spent years catching up.

AI is in that phase right now. Most organizations use it as a thin layer over the existing work: draft this email, summarize this document, generate this code. Useful, and nowhere near the whole story. The bigger question is how AI reshapes service, research, development, compliance, procurement, sales, onboarding, and decision-making, and most organizations have not really asked it yet.

Privacy rules, cybersecurity frameworks, platform governance, digital identity, content moderation, consumer protection: all of it arrived after internet adoption was already widespread. Governance lagged, and society paid for the gap in the meantime.

AI is moving through the same gap, only faster. Organizations cannot wait for perfect regulation before acting, because adoption is not waiting for them. But they also cannot let every employee use any tool with any data for any purpose, because that is exactly how the gap turns into incidents.

So they need internal rules now: what data can be used, which tools are approved, where human review is required, how outputs get verified, how incidents are escalated, and who is on the hook. Not a finished regulatory regime, just enough governance to make safe use easy and risky use visible while the external rules catch up.

The internet turned out to be far more than a communications channel. It reorganized discovery, distribution, commerce, media, and social life. AI will outgrow the productivity-assistant label the same way, by becoming part of how work is structured instead of a tool parked next to the work.

The mistake to avoid is treating this early, crude phase as the technology's permanent shape. Judging AI by today's draft-an-email use case is like judging the internet by the brochure website. Most of the interesting part is still ahead.

Keep both halves of what the skeptics got partly right. AI will spread, probably faster than anything before it. And uncontrolled spread will create real problems that thoughtful governance can prevent or contain. Plan for both at once, and you are already ahead of where most organizations stood through the internet's first decade.