By Ravi Gangal and Shruti Aji Murali
We are all familiar with rapid, parallel price increases closer to the date of travel, and even parallel “surge” pricing with cabs during peak traffic. While these are examples of algorithms responding to changing customer demand - they may also seem like an indication of competition gone wrong. Industries like airlines, hospitality, and cab-hailing commonly use AI tools to determine prices, manage inventories, and allocate resources because they can easily track data and set prices in real time - often resulting in similar pricing trends.
AI is expected to become more cutting-edge and autonomous in the near future. While automation has obvious cost efficiencies, competition authorities are looking into the possibility that these technologies can disrupt competition and harm consumers.
One area of concern is how AI can be used to facilitate cartels. Dynamic pricing models can be opaque, making it difficult to determine whether collusion has taken place and who is responsible for it. These programmes can quickly process large volumes of data, enabling companies to respond more rapidly to changing market conditions. As such, regulators across the world continue to assess AI’s evolution, capabilities and applications, to ensure that it is not used for market distortions.
Algorithmic collusion and accountability
AI enabled algorithms operate on a vast spectrum of autonomy. Some algorithms strictly adhere to pre-set guardrails, where the scope of human intervention is clear. Others, particularly those being developed, are more autonomous, and are guided by a broad objective - here, the human intervention is more limited. An algorithm’s autonomy depends on its specific design, the data it uses, and the market environment in which it operates. With greater autonomy, outcomes become less predictable, blurring the lines between human and machine actions.
This has led to various jurisdictions assessing if AI could be a cartel enabler and if some industries have already begun demonstrating this trend.
Between 2014-21, the Competition Commission of India (CCI) scrutinised the use of dynamic pricing software in the Indian airline sector, but did not find any collusion. Although the airlines used similar software programmes, each had revenue management personnel who independently determined the final fares.
In the EU, several travel agents used a common booking platform to set prices, and were held liable for colluding through the platform with respect to a discount cap. Here, even companies that were aware of collusion taking place through the platform, while not actively participating were held liable. Importantly, it was noted that companies had not actively distanced themselves or reported collusion to the regulator.
There are also ongoing cases in the US, which relate to pricing algorithms used to set lease rental rates. In these cases, landlords shared confidential rental data to train the pricing software, which then recommended prices, reducing independent competition and potentially inflating rents.
As more regulators join this trend, companies may have some learnings in these cases - first, whether their AI software programmes merely recommend prices and prices are independently set or these programmes facilitate collusive price fixing (the latter is no different from a hub-and-spoke cartel where the algorithm is merely the conduit - something the CCI assessed and rejected in the Cab Hailing Case (2018); second, the nature of the information companies share with these programmes - the more competitively sensitive the information, the more likely that collusive outcomes occur; third, antitrust liability can also extend to even passive cartel participants - who are aware of cartelisation taking place and do not actively distance themselves.
But what if companies have no idea that the software that they use for efficiency gains - also shares their confidential information or uses it to churn out results for its other users? Can companies still be penalised? Alternatively, would it be possible to hold the algorithm or it’s creators, liable for collusive outcomes? These are some questions that the regulators must account for, before they decide on AI’s fate.
Where does the CCI go from here?
The CCI’s AI market study is timely. Understanding AI’s nuances is crucial to identifying emerging competition concerns, especially when India’s AI ecosystem is nascent. CCI market studies are also a useful signal to the market to clarify its enforcement priorities.
As AI tools evolve, the CCI must continue to adopt a case-by-case approach to hold parties accountable where AI is used intentionally to collude, causing harmful market distortions. It is essential that the CCI’s intervention does not unintentionally stifle businesses from using AI legitimately to enhance their competitiveness through innovation.
(The authors, Ravi Gangal and Shruti Aji Murali, are competition lawyers at Axiom5 Law Chambers LLP.)
Views are personal, and do not represent the stance of this publication.
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
