For a few months in 2001, the business world couldn’t stop talking about inventor Dean Kamen’s secret project, codenamed “Ginger”. Kamen had received the National Medal of Technology from US president Bill Clinton in 2000 for his groundbreaking work on medical devices, including the first portable insulin pumps and dialysis machines, and the technorati were sure Ginger would be used by everyone.
After getting a preview, Steve Jobs said Ginger would be “as big a deal as the PC”. John Doerr, whose venture capital firm Kleiner Perkins had backed Amazon, declared it might be bigger than the internet, and put in $38m—at the time Kleiner’s largest investment in any company. The Ginger hype was such that South Park dedicated an episode to mocking it.
After all that, the big secret turned out to be the Segway: the admittedly cool scooter-like device that allows individuals to travel intermediate distances at up to 12mph. The technical achievement—outsourcing to an array of sensors, motors and machine intelligence the human ability to balance while standing up—might well have justified the $100m Kamen marshalled into its development, but Segways haven’t transformed lives, other than those of local entrepreneurs who use them to conduct city-centre tours.
The same can’t be said, of course, of machine learning, the technology behind the software that both stabilises the Segway and powers AI chatbots. It is transforming virtually every industry, but comes with relentless noise. With some warning of a disastrous bubble in AI stocks, and machine intelligence pioneer Yann LeCun warning that companies like OpenAI are marching into a dead end, we need a better way to assess the visions hyped by AI pioneers.
Meta is possibly the best example of a company trying to shape reality. In 2021, Mark Zuckerberg was so excited about the Metaverse, a virtual fantasyland where crude, legless avatars mingled, that he rebranded his company after it. Meta’s logo resembles a VR headset, a technology that Zuckerberg promised would transform first the internet and then all society.
You are almost certainly not reading this column on your Meta headset: with all the company’s marketing muscle, it sold only 20m headsets worldwide between 2021 and 2023. Early this year, in a “pivot to AI”, Meta closed three VR game studios and laid off 1,000 employees focused on virtual reality. It’s possible Meta rebranded at a bad time, just as Covid lockdowns ended and people were desperate to interact in person. But it’s also possible that the Metaverse is as dumb an idea now as it was in 1995—when the first internet companies discovered that VR would be a niche business at best.
But at least overhyping VR didn’t hurt anyone, unless your VR headset gave you nausea. You can’t say the same of cryptocurrency. The steep transaction costs quickly made it clear that digital money was not going to provide a decentralised micropayment system. Rebranded as “digital gold”, crypto was then sold by acolytes as an asset that, like a precious metal, would retain its value when banking systems inevitably collapsed.
Frustratingly for crypto-sceptics like me, some currencies have kept their value, though not without stomach-churning volatility. But millions of other projects have collapsed in value, proven to be scams or otherwise financially ruined their investors.
Crypto has generated a whole ecosystem of ways to lose money to grifters—NFTs, memecoins—and novel ways to enable illicit financial transactions or bribery. Donald Trump’s family has shown how these phenomena go together. The New York Times estimates that the Trumps have netted more than $850m in various cryptocurrency deals, often with businesses that wish to curry favour with the US government.
Like the Segway and the Metaverse, cryptocurrencies generally function as intended. But they have not transformed society, or certainly not for the better. Many of today’s AI visions—cars without human drivers, companies run by a single human commanding an army of robots and AI agents—are precisely the kind we should be sceptical of: futures desirable to technologists and investors, but perhaps not to society more broadly.
Tech hype relies on “technodeterminism”—the idea that a technology, once conceived of and sufficiently developed, will spread throughout society due to the efficiencies it creates. While this is a reductionist theory it’s not without merit: innovations like rail, flight and the internet have all had a hand in major societal transformations, displacing existing institutions and practices. Even apparently small advances—like container shipping—can thoroughly disrupt major systems such as global manufacturing and trade.
Technodeterminism is also deeply appealing to the technorati. It tells them when they back a new product they are not merely solving a problem and making a buck: they are working to better society. But its greatest trick is creating a sense or narrative of inevitability: it leads us to think a certain future is coming, whether we like it or not.
In 1995, consulting firm the Gartner Group introduced the “Hype Cycle of Emerging Technologies”, a compelling and simple graph that highlights a predictable pattern: when a new technology first appears, expectations are massively inflated; this is followed by a period of disillusionment and then finally a “plateau of productivity”, as the technology in question finds its true place in the market and public life. A version of the graph is released annually that places technologies in the news at precise locations on the graph’s curve.
The 2024 graph—the most recent one available for free—warns “generative AI” is entering the trough of disillusionment, but should reach the plateau in two to five years, while autonomous agents are still climbing the peak of inflated expectations.
On the one hand, the graph warns us that even the most transformative technologies tend to be overhyped at first. But it’s easy to misread it as telling us all tech will eventually be adopted. Tech companies love this interpretation: their failures aren’t evidence of poorly conceived or badly implemented products—they just haven’t reached their plateau of productivity yet.
The Gartner graph is helpful only if we remember that while many technologies transform societies, not all do. The Segway can cover ground quickly, but many people (myself included) find it counterintuitive and scary to use. That barrier to adoption, combined with cost, weight and the challenges of the built environment, have made it a niche product useful for a few applications, such as policing airport terminals and other stair-free environments.
Even trickier than designing vehicles that can negotiate stairs is the task of winning social acceptance. People truly hated the idea of their lives being recorded by users of Google Glass, the smart glasses that allowed wearers to add a data display layer to the world they moved through. While the tech was deeply impressive, public reaction to being put under surveillance made wearing Google Glass deeply uncool.
Meta, a company that’s perhaps one of the least sensitive to what users will tolerate, is now peddling new smart glasses using frames from Ray-Ban. The next few years will show whether fashion trumps social embarrassment.
It’s no mystery why John Doerr would hype the Segway: he had money riding on it and, if he were right, he would be considered a seer and shaper of futures. The firm he worked for was thoroughly insulated, too. Venture capital is based on the idea that most investments fail, but the great successes outweigh the costs of the failures. This also means that, at the very top, there are often few consequences for getting predictions wrong.
As consumers and as a wider society, we need to grow wise to the hype cycle of failing technologies. We need prophets who aren’t motivated by profits, nor afraid to say that while some tech will move through the stages of hype and disappointment into everyday use, lots of others will simply die on the vine. And still more we may, as a society, choose to kill off.
