Low productivity and rising prices in healthcare, construction and education could finally be disrupted by recent technological applications. Moreover, artificial intelligence is expected to improve healthcare far more than ICT ever could. However, concerns remain whether greater welfare in these sectors will lead to measurable productivity increases.
Observations
- In The Rise and Fall of American Growth, Robert Gordon argues that technological innovation is very unlikely to ever again make productivity reach the levels of the special century (1870-1970).
- Marc Andreessen argues that only some industries (healthcare, construction, education) suffer from low productivity growth: a lack of technological adoption is to blame for rising prices in these sectors.
- Recent technological applications could lower prices in these industries. Prefabricated homes and 3D printing could raise construction productivity; robots could raise productivity in health care; online learning may come to offer free high quality education to a global audience.
- An AI arms race among the Big Five promises even more improvement in low productivity industries, especially in healthcare. Deep learning, a subset of AI, is already focused on the medical realm. A lot of what doctors do is image recognition, while deep learning systems promise greater accuracy and faster analysis.
- Deep learning systems have taken off in recent years. The capacity of their ‘neural networks’ has increased exponentially. Moreover, deep learning systems can already describe what they see in images and researchers have reached human parity in speech recognition.
Analysis
Recent technological applications in healthcare, construction and education promise greater welfare and lower prices in sectors that have suffered from rising prices for decades. Erik Brynjolfsson argues that the productivity revolution with computers as a general purpose technology (GPT) has only just begun. Indeed, only recently have Uber and Airbnb brought dramatic improvement to two large industries that have been more or less stuck for decades. It seems that Gordon’s vision could be too pessimistic: GPTs of the past have always taken decades to realize their full potential.
The productivity revolution with computers as a general purpose technology (GPT) has only just begun
AI already promises to improve healthcare in a way that traditional computers and ICT have not been able to do. While Brynjolfsson argues that we are on the brink of a productivity boom with computers as a GPT, Andrew Ng argues that ‘AI is the new electricity’. The Big Five believe that AI will improve every sector of the economy, including those with low productivity growth. Indeed, deep learning is already focused on the medical realm: with far better data management, cognitive assistance and image recognition, AI and deep learning are able to significantly improve healthcare. The only hurdle that remains is low technological adoption (primarily due to regulations) in low productivity sectors like healthcare and education.
However, these productivity revolutions may after all not lead to measurable increases in productivity and economic growth, even though they will increase welfare and lower prices for consumers. Indeed, on that account Gordon may have been right: he was worried about the end of economic growth. The concept of GDP is increasingly under pressure. In a world where houses are Airbnb hotels, cars are Uber taxis and Youtube brings hours of daily entertainment to hundreds of millions at no cost, GDP is increasingly a misleading measure. Therefore, new technological applications could increasingly improve people’s lives, while the economy stagnates.
