Joshua Gans(トロント大)が表題の点を追究したNBER論文を2編上げている。一つは「AI Adoption in a Competitive Market」(ungated[SSRN]版)で、その要旨の冒頭でGansは

Economists have often viewed the adoption of artificial intelligence (AI) as a standard process innovation where we expect that efficiency will drive adoption in competitive markets.


The tendency in the economic analysis of AI has been to equate AI adoption with automation. While there are applications of AI that are embodied in capital, AI is fundamentally an improvement in prediction technology and, therefore, its first order impacts will be to improve decision-making under uncertainty. When considering, therefore, the impact of its adoption of different classes of factors of production, AI adoption (say as measured by the employment of people with AI skills) will be a complement to inputs that are more variable in the short-run. AI prediction allows those inputs to be chosen to better respond to changes in variables and hence, increases their efficiency on average. If such variable inputs are in terms of labour and its employment, AI adoption will be primarily labour augmenting.
This paper has examined AI adoption from this perspective in a competitive market. Even here the adoption of AI has external effects on other firms. While most efficiency-enhancing innovations would involve adoption that reduced the profits of competing firms, here it is possible that AI adoption could increase the profits of competing firms. This may, in turn, limit the adoption of AI across such markets. It is shown that this arises, however, when variable inputs impacted on by AI are a smaller share of total inputs used by the firm. The broader AI’s impact within a firm, the stronger will be the incentives to adopt and these will be driven by competition. This suggests that researchers will need to be careful in measuring the adoption and impact of AI in such markets.
There are many other issues that can be explored regarding the adoption of AI. One such avenue is to consider AI adoption by firms with market power – that is, who do not take market prices as given. This is explored by Gans (2022) who finds that the value of AI adoption differs upon whether it is informing pricing, output choices or both.

ここで参照されている「ガンズ(2022)」がもう一本のNBER論文「AI Adoption in a Monopoly Market」(ungated[SSRN]版)である。以下はその要旨。

The adoption of artificial intelligence (AI) prediction of demand by a monopolist firm is examined. It is shown that, in the absence of AI prediction, firms face complex trade-offs in setting price and quantity ahead of demand that impact on the returns of AI adoption. Different industrial environments with differing flexibility of prices and/or quantity ex post, also impact on AI returns as does the time horizon of AI prediction. While AI has positive benefits for firms in terms of profitability, its impact on average price and quantity, as well as consumer welfare, is more nuanced and critically dependent on environmental characteristics.