An article published by the Financial Times
The second quarter US gross domestic product growth due to be announced on Friday morning is to be the highest since it hit 5.2 per cent four years ago. Does this mean we should stop worrying about low economic and productivity growth?
Yes. But not because of the numbers. GDP growth bounces around a lot: it was just 2.6 per cent in the first quarter. What’s more, about half a per cent of the current growth is likely to be due to the temporary stimulus of the deficit-financed tax bill. That is not sustainable.
The cause for optimism is more fundamental. My research with my colleague at Massachusetts Institute of Technology, Andrew McAfee, shows that recent advances in machine learning, a form of artificial intelligence, will fuel a productivity and growth surge — not just in the US but around the world.
We have heard a lot of hype about AI, but the reality is that the field has finally begun to deliver on the promises it first made more than 60 years ago. Machine-learning systems have recently demonstrated superhuman performance in domains as diverse as recognising objects in images, detecting fraud, diagnosing disease, making recommendations to customers, and playing poker.
Consider the annual global competition for machine labelling of images. In just eight years, the error rates have fallen from 28 per cent to 2.5 per cent, and the winner can now beat the 5 per cent error rate achieved by humans.
Last year, ML systems matched board-certified dermatologists in recognising skin cancer from images, and they can now diagnose many other diseases from medical images. The voice recognition systems we increasingly use in our homes and smartphones are powered by ML. These systems are far from perfect, but we are in a historical transition from a period when machines could not understand us when we spoke, to one where they can answer our questions and carry out simple instructions.
In a paper with Daniel Rock and Chad Syverson, we discuss how machine learning is an example of a “general-purpose technology”. These are innovations so profound that they trigger cascades of complementary innovations, accelerating the march of progress and growth — for example, the steam engine and electricity. When a GPT comes along, past performance is no longer a good guide to the future.
If machine learning is already superhuman, why did we not see it in productivity statistics earlier? The answer is that its breakthroughs haven’t yet had time to deeply change how work gets done in call centres, hospitals, banks, utilities, supermarkets, trucking fleets, logistics management and other businesses. Yes, predictions about click-rates already power the placement of ads on social media, but nearly 50 per cent of food is still wasted on its way from farm to table because we do a terrible job of predicting supply chain snafus.
Technology is a catalyst, but technology alone will not bring a productivity boom. Entrepreneurs need to invent new business models, workers need to develop new skills, policymakers need to update rules and regulations. What is more, they can do it in ways that create shared prosperity.
The Inclusive Innovation Challenge at MIT has been set up to help speed the transition to a high-growth, high-opportunity economy. As ML systems develop, we can have not only higher productivity growth, but also more widely shared prosperity. The productivity boom is waiting to be unleashed.
Erik Brynjolfsson is a professor at MIT and co-author of ‘The Second Machine Age’ with Andrew McAfee, who also contributed