Artificial Intelligence (AI) promises to bring significant procompetitive consumer benefits but these developments are accompanied by a number of potential risks relating to competition. The OECD’s James Mancini highlights some of these risks and looks at competition policy options that can help to ensure that AI reaches its procompetitive potential.
Artificial intelligence has the potential to reshape how decisions are made in markets. This can mean significant benefits for consumers in the form of new products, lower prices from more efficient business processes, and even assistance in making complex purchasing decisions. However, in reshaping market dynamics, AI could also seriously dampen competition in ways that may not be easily addressed using existing competition enforcement tools.
Imagine a market where AI is given free rein by human managers to set prices and make other product decisions using a rich set of market data. Sophisticated AI applications may, depending on the objective they are given, decide that collusion with competitors is an optimal outcome (for example, in order to maximise a future stream of profits). Thus, price wars and aggressive competition by risk-taking human managers may be replaced by stable and high prices as well as poor quality products. To reach such an outcome, different firms’ AI may use signalling strategies to indirectly communicate with one another, and thus jointly make decisions on prices or other variables. The risk of this outcome would be particularly high if firms used the same AI tool from a third-party provider, or if they had access to an identical flow of market data.
The potential for such an outcome has generated a great deal of speculation and concern. AI in this scenario would make collusion easier to implement, more durable, harder for competition authorities to detect, and potentially even fall outside the scope of competition laws.
While the example of AI reaching a collusive outcome completely independent of any human role is an extreme one, there are other risks as well. Rather than colluding with competitors, an AI decision-maker could select aggressive strategies that cross the line into conduct considered to be an abuse of dominance or attempt to monopolise the market (depending on the jurisdiction).
AI can also be used as a tool for implementing anticompetitive strategies developed by humans, whether they be collusive agreements or abusive strategies to exclude competitors from markets. Concentration in AI capacity, and the data needed to make use of it, could also give some firms durable market power that cannot be easily contested. Mergers involving firms with substantial AI capacity and data access may therefore need to be scrutinised carefully.
Whether these risks manifest remains to be seen – there have only been a few cases of collusion being facilitated by automation, and these involved only simple pricing algorithms. However, competition policy makers may need to consider changes in order to ensure they are ready to confront AI competition problems. This could include legislative changes, to ensure anticompetitive conduct by AI is captured. In addition, several jurisdictions are investing in greater technical capacity to be able to assess AI, and even potentially use AI for investigative purposes. Finally, competition authorities may need to make use of advocacy tools and market studies to help tackle the underlying conditions that give rise to AI-related competition risks.
In sum, it is still too early to say whether AI will deliver on its potential for significant procompetitive consumer benefits, or whether it will lead to widespread competition harm. However, it is clear that competition policy will have an important role to play in managing the potential dark sides of AI technology for consumers.
The 2021 OECD Business and Finance Outlook, scheduled for release on 24 September 2021, focuses on AI in Business and Finance. Chapter 4 of this publication explores in depth some of the potential competition risks stemming from the use of AI, namely collusion and abuses of dominance, and highlights the challenges they pose for competition policy.