AI Vey (Shared as a Public Service.)
The Reductionist is really growing to love The Economist, and not just for the famous ads-in-the-red-box they’ve put out over the years. Since deciding to pay the paywall piper, I’ve been consistently impressed by the thinking, the “there’s more to the damned planet than just US” insights, and, yeah, the writing. In that spirit, here’s a goody excepted from this morning’s highly relevant missal:
“In 1987 Robert Solow, a Nobel-prize-winning economist, famously quipped that you could see the computer age everywhere but in the productivity statistics. Although a productivity bump did, in fact, emerge in the late 1990s and early 2000s, it was not particularly large. The uplift hardly took Western economies beyond their typical growth rate of around 2% a year.
Now the burning question is whether artificial intelligence could be different. That is what I have been asking economists at this week’s National Bureau of Economic Research’s summer institute—an annual gathering for the economics profession, which is taking place in Cambridge, Massachusetts. So far, nearly everyone has responded with a version of: “Possibly, but it will take longer than you might think.”
Productivity improvements rely on technological “diffusion”—the spread of new capabilities across business and geographies—which does not always follow from innovation. And there is a worrying trend: diffusion may have actually been getting slower in the past two decades.
The idea sounds counterintuitive. Surely modern innovations, such as Google and social media, should make ideas whizz around faster? Wasn’t ChatGPT the fastest-growing consumer application in history?* Yet as my colleague, Callum Williams, and I write this week, there is a big difference between consumers adopting new technologies like AirPods or smartphone apps and the full-scale reorganisation of businesses around new technologies.
Bosses of firms routinely tell us of the challenges they face with AI. Some have not been saving their internal data in a way compatible with AI models. Others are afraid to send data off to an AI firm for security or liability reasons. According to one survey, around 40% of small firms in America are not even trying to use AI.”
Headline is “Money Talks: How AI fails.” Author is Arjun Ramani, the pub’s global business and economics correspondent and you can look it up for the full scoop. Worth reading.
*Reductionist note: actually, Threads is now the fastest growing consumer app in history. Buy a whole bunch.