The Big Tech hype cycle often goes nowhere

Those of us still gainfully employed in white-collar jobs may be permitted a moment of self-congratulation. Two years ago, the launch of OpenAI’s ChatGPT was accompanied by dire predictions that “knowledge workers”, as we are somewhat pretentiously known, would be the first victims of the next industrial revolution. Our pathetic emails and data analysis skills were to be automated.

OpenAI’s CEO Sam Altman — grave-faced and clad in the smart suit Silicon Valley’s titans don only at moments of utmost crisis — toured the parliaments of the world warning politicians of the imminent necessity of a universal basic income to support their stricken workforces through the coming age of mass unemployment.

The past 18 months have failed to live up to Altman’s rather gloatingly apocalyptic predictions. A recent Goldman Sachs report wonders whether two years of massive AI investment “will ever pay off”. The much-promised GPT-5 has failed to materialise. The current version’s technology is impressive but far from omnicompetent. Its maths is execrable and its reasoning eccentric. I recently spent a diverting evening trying to convince ChatGPT that the word “strawberry” contains three “r’s. (“Yes, I’m 100 per cent sure. The word ‘strawberry’ has exactly two ‘r’s.” Its crazier solecisms — such as the advice that “pebbles, geodes or gravel” may form part of a balanced diet — are now items of internet lore.

Generative AI may yet develop into human-level intelligence. And it may not. Many people have been too quick to assume the former without any specially good reason. Trust in the inevitability of scientific progress runs deep in our culture. A half-century of wonders has accustomed us to think of technological innovation not as a fallible human endeavour but as magic via circuit boards. With enough persistence and money, the frontiers of science must advance. Nowhere is this faith stronger than in Silicon Valley, where every tech investor’s career is predicated on the belief that the “next Facebook” is out there somewhere, requiring only a suitable injection of cash to balloon up to society-toppling proportions.

• How AI firms take on ‘niche and boring’ work to free up humans

Too often neglected is a counter-history of innovation: a story of plateaus and cul-de-sacs. Jet-powered flight has not improved in decades; since Concorde was decommissioned, it has arguably gone backwards. The effectiveness of antibiotics is degrading as bacteria develop resistance. The promise that we are on the brink of limitless energy through nuclear fusion is more than half a century old.

There is a case that fundamental physics has seen no truly important breakthrough since the 1970s. Our counter-history’s chapter on Silicon Valley might cynically propose that one genuine revolution — the invention of the iPhone — has been succeeded by various clownish attempts to force another paradigm shift that won’t arrive: Google Glass, the metaverse, bitcoin. Elon Musk assured us that fully self-driving cars were imminent in 2014 and has repeated that vain promise almost every year since.

This is not to say technology never improves, only that progress is far from inevitable. Not all scorned tech visionaries are latter-day Galileos facing down the papal court. Some are deluded optimists.

A recent analysis in The Economist usefully debunked the modish theory of the “hype cycle” (popular, unsurprisingly, in AI circles) which states that every transformative technology begins by attracting enthusiasm and investor cash, then enters a period of latency in which money and enthusiasm leak away while everyone tries to work out what it is for, until it is triumphantly resurrected when some bright spark figures out the lucrative-use case and reaps the rewards of the initial investment.

• A third of managers in the UK have never used AI

The classic instance is the dotcom bubble, which ballooned then burst before the internet at last began its final ascent, partly enabled by the infrastructure of fibre optic cables built at the height of digital enthusiasm in the late 1990s. The Economist found that this trajectory was rare. Plenty of technologies attract hype, then crash, then go nowhere: carbon tubes, household 3D printers, the radically decentralised “Web3”.

Credulity about scientific progress suits tech founders hungry for money. And it conveniently licenses bad behaviour. Mark Zuckerberg’s motto “move fast and break things” implies trampled laws and broken copyright are necessary sacrifices to a better future. “Sit still and break things” doesn’t carry the same moral conviction. “AI” now provides useful PR cover for any industry trying to make old-fashioned cost-cutting seem sexily future-focused.

Meanwhile, media brutalised by two decades of technological innovation — alas for newsprint! — are accustomed to assuming the worst. And few journalists are keen to be remembered in the same breath as the researchers quoted in the newspaper article of December 2000 headlined “Internet ‘may be just a passing fad as millions give up on it’”. Somehow, credulity seems more hard-headed. Hence Silicon Valley bustles with overpromising hucksters and charlatans. Sam Bankman-Fried and Elizabeth Holmes were beneficiaries of our naive faith in technology.

This is not to say that generative AI is incapable of fulfilling the grand promises made for it — I am not qualified to judge — only to argue that the sceptical case should be heard, too. Facebook’s head of AI, Yann LeCun, is one of many experts who argues that generative AI is too flawed to ever develop humanlike intelligence. Technology is not magic. Progress is not always guaranteed. Perhaps this — in our science-worshipping age — is the really disconcerting thought.