What does AI mean for productivity?
A Dall-e image of an economist reflecting on the impact of AI on productivity.
The media likes to pitch the bullish technologists against the bearish economists.
This dichotomy is unhelpful.
Many, if not most, economists have come to agree that AI will lead to some cost savings and productivity improvements, given its transformative potential in a wide range of economic activities.
It is considered as the latest general-purpose technology (GPT), similar to previous digital technologies such as computers and the internet.
The question economists seem to *disagree* on is around the size, timeline and type (net positive or negative for jobs) of AI's impact.
These new perspectives are gathering more attention.
💡 Transforming innovation, not just production.
Most economic analysis focuses on automation, which is linked to the production processes in an economy. But new work focuses on deep learning, which is linked to the innovation processes in an economy. The reason? Deep learning has been the driving force behind numerous groundbreaking achievements over the past decade.
This is the premise of work by Tamay Besiroglu et al. They argue this is the way AI will accelerate productivity, by boosting scientific knowledge and the ability of R&D to solve problems. The authors also use machine learning to estimate human capital, indicating that the very field of economics could be revolutionised by AI.
🏢 Tech must go hand-in-hand with organizational change.
This is encapsulated by the idea by Erik Brynjolfsson et al. that when a new GPT is introduced, it requires significant complementary investments by businesses to capture the productivity benefits. Complementary investments include creating new business processes and business models, invention of new products and investments in training management and people.
All of these are intangibles, which are poorly measured in economic statistics, and make it hard to find quant evidence of the impact, initially. But over time, the impact gets reflected in the productivity data. This means we should expect a J-shaped productivity curve. There is a long way to go: less than 5% of firms reporting the use of this technology in the US.
As I was reading various papers I could not help notice a gap in the research of female economists on technology issues.
The work of Ester Duflo, Anne Case and Mariana Mazzucato has revolutionised our thinking in their respective focus areas.
But I've not seen this in the tech space yet, when clearly different perspectives would be enormously useful in what Max Tegmark calls 'the most important conversation of our time'.
Given how intertwined technology is with other issues our world is grappling with, such as climate change and inequality, where are their voices? And how might the narrative of economics on AI and generative AI look different if their perspectives were more visible?
