Will AI Change the Cost of Everything?
At a time when supply chain disruptions, labor market shortages, and monetary policies have driven up the cost of goods/services, we are not talking about AI’s likely disinflationary impact
How AI will impact jobs is getting a lot of attention. For good reasons. But there is another big economic impact we aren’t discussing at all. I believe that AI will also drive an increase in consumer affordability. And it will affect different wallets differently—let’s dig into that.
AI will impact the cost structure of goods and services through several mechanisms. The first is automation. AI will displace humans for a range of tasks because it can get those tasks done at a lower cost; a lower cost of production can mean higher margins for the producer, a lower price tag for the consumer, or – especially in competitive markets – both. For example, automation during the second half of the nineteenth century dramatically lowered price per yarn of cloth and increased affordability of clothes and other textile items (also caused the Luddite revolt against technology).
The second is augmentation; AI does not just replace workers, it also acts as a copilot and boosts their productivity, by taking on some tasks to clear their plate for others. For example, a video editor armed with AI might soon be able to edit a movie in weeks instead of months. In general, as labor becomes more productive, wages rise. How much wages rise, and for whom depends on which workers experience the greatest productivity gains. Studies suggest Generative AI lifts the output of lower-skilled workers more than for higher-skilled workers for a range of knowledge work. One study showed that an AI copilot allowed developers to complete software development tasks 55.8% faster but developers with less programming experience benefited the most. This narrowing of the productivity gap reduces higher-skilled workers’ bargaining power and wage premium. This in turn suggests any increase in wages from augmentation will be driven from the bottom end of the wage spectrum among knowledge workers. What will this do to prices? Despite increase in wages, increase in productivity usually expands supply and lowers prices for consumers.
Third, AI lowers entry barriers across many industries, enhancing competition and leading to lower prices. In enterprise software, for example, AI is drastically reducing development costs, enabling startups with limited resources to build enterprise-grade products. My start-up, which applies AI’s human-augmenting powers in Hollywood, wouldn’t exist without AI, since I could not have afforded the order of magnitude more software developers we would have required if we had built our tool even five years earlier.
To be clear, AI's disinflationary – and, in some cases, deflationary -- impact will vary across sectors. Software is likely to see price declines, but the real opportunity lies in reining in education and healthcare costs. For example, U.S. national healthcare expenditure per capita went up by 311% on an inflation-adjusted basis during the period 1980-2020 and college tuition went up by 181% during the same period. Lowering of these costs through a combination of automation, augmentation, and lower entry barriers will go a long way in increasing affordability. But in sectors like energy, AI might drive up prices given the energy needs of today’s AI.
Economists worry that long-term deflation could contract economic activity if people wait for ever-lower prices. I don’t think AI will trigger lulls in buying. Falling prices can also stimulate demand, as evidenced by the television industry's growth as TV prices dropped over the last several decades. New technologies also spur new uses for capital and new tasks for labor. Companies are already assessing which parts of their products and processes will be made obsolete by AI and investing in their transformation. And let’s not discount human ingenuity -- our ability to leverage AI to create entirely new product markets and employ labor in new ways. As the cost of recording and projecting moving pictures dropped in the early 20th century, it gave birth to Hollywood; as the cost of sharing photos and videos further plummeted in recent years, it gave birth to social platforms like YouTube and Instagram and the creator economy. AI will likely help create whole new markets even if its early impact appears to center on driving operational efficiencies. There is no doubt that AI will change the basket of products we consume and it’s hard to speculate on how the new products will be priced. Regardless, such shifts will be driven in part by the lower prices of the goods we currently consume.
In summary, while the labor market risks posed by AI are real – particularly if its impact is more automation than augmentation – its disinflationary impact also needs attention. How much lower the prices go will depend on how competitive the market is and how accessible the productivity-enhancing AI is to all firms in the market. A publicly and cheaply available LLM that helps people code might reduce prices for website design; a proprietary LLM tapping private datasets of, say, market movements will not necessarily drive down fund management fees.
Very insightful since we dont generally think about the deflationary angle when talking about Gen AI. I wonder how this will impact the pricing models of enterprise SW and HW.
But Professor no body is sure if AI displaces labor market what will happen to consumption so even if AI lowered the prices there will be no demand for those products.