Artificial Intelligence (AI) and big data are playing an increasingly important role in the economy. Competition law will have to evolve if it is to reconcile the positive and negative effects that the use of these new instruments may have on markets.
Modern life is being progressively influenced by the “fourth industrial revolution”, whereby new and disruptive technologies and trends, such as the Internet of Things, robotics, virtual reality, AI, and big data are combined in ways that until recently were inconceivable. Although all these factors will influence the way we work and live, AI and big data are already playing a critical role in how companies make business decisions and interact with customers, suppliers, and one another. AI may even lead to the development of entirely new economic models.
Because of the relative novelty and rapid development of AI-based platforms and applications, antitrust agencies and commentators are struggling to establish how the existing competition law framework can be applied to the issues that arise out of the effects that AI and its use of data have on the functioning of markets.
AI is based on a tailor-made set of algorithms. An algorithm is itself a set of decision-making rules that is able to generate a precise output from certain data inputs. Traditional algorithms are relatively static (if this then that) and do not adapt their output based on feedback. AI introduces a self-learning element that allows the algorithm to adapt itself and get better and more efficient at its task as it processes more data. AI is therefore a dynamic algorithm with an increasingly precise output.
AI only works, however, if the underlying algorithms are fed sufficient data that allows the algorithm to learn and improve. For example, when a consumer searches online for a product, a supplier’s AI platform will collect data from the customer’s previous searches, and combine this with a detailed profile of the individual and statistics from the market, in order to propose a new product at the right time, personalised and priced appropriately to entice the consumer to make the purchase.
If the consumer buys a different product through the platform, that choice will become part of the updated dataset that determines what product may be proposed to this or similar consumers in the future. The chances that the proposed offer appeals to this consumer therefore increases over time.
AI, algorithms, and access to data clearly influence the functioning of markets and may therefore have direct or indirect implications under competition law.
Collusion in oligopoly markets is one of the main concerns of competition agencies when assessing the effects of algorithms, even though enforcers and academic commentators differ on whether the use of algorithms will ultimately harm or benefit competition.
The use of AI in pricing algorithms is often feared to lead to tacit collusion i.e., the alignment of competitive behaviour without explicit agreement. Because algorithms are created with a specific purpose in mind, their propensity to result in tacit collusion depends on how the algorithm is programmed and used. Much depends on the number of competitors in the market and the willingness by some to compete aggressively.
Equally, it has often been argued that pricing algorithms create more opportunities for consumers to compare products and prices and ultimately get a better deal. In a market with sufficient suppliers, this is likely to hold true. In a market with few suppliers, this transparency may have the opposite effect and induce these few suppliers to align their prices.
RESALE PRICE MAINTENANCE
The first fine imposed by the European Commission in this field related to the use of algorithms in e-commerce and concerned a vertical issue. The Commission sanctioned four consumer electronics manufacturers that had engaged in resale price maintenance (RPM) practices with regard to their online retailers.
When those retailers did not follow the prices requested by the suppliers and instead offered their products at prices below the recommended resale price, they faced threats or sanctions from the manufacturers, such as blocking of supplies. The use of sophisticated monitoring tools allowed the manufacturers to effectively track resale price setting in the distribution network, and to intervene swiftly when it noted deviations from the recommended resale price. The price interventions limited effective price competition between retailers and led to higher prices for consumers.
An interesting element here is that the retailers themselves used pricing algorithms, which automatically adapted retail prices to those of competitors. In the Commission’s view this would have led to lower prices if it were not for the intervention by the manufacturers.
In addition to these collusive aspects, it is clear that algorithm-based business models could also give rise to potential abuse of market dominance by large operators.
ABUSE OF DOMINANCE
By using algorithms and other new technologies, dominant market players may find new ways of leveraging their dominance and foreclosing other companies from the market. This was the case with a large search engine provider that was found to have abused its dominant position by altering, for its own benefit, the criteria of its generic search algorithm with the objective of demoting competing comparison shopping services in the search results list.
A main area of discussion regarding dominant companies is whether or not they should make their data sets available to companies that, without this data, will never make it in a given market. Access to data is therefore likely to be a hot topic for the new Commission.
Algorithms might entirely change the way suppliers interact with their customers in a market. By monitoring prices, customer profiles and behaviour, and other factors, algorithms can create the “perfect price discrimination”, which is the ability to charge customers exactly what they are willing to pay at any given time and circumstance. This individualised pricing is still relatively new from a competition law perspective, and raises the question of whether or not offers and transactions can be compared, and how they should be assessed from a competition law perspective.
On the other hand, consumers will also increasingly have access to platforms that use pricing algorithms to make markets more transparent and navigable, and help consumers make better choices. The same platforms may allow producers to react more quickly to consumer demand and market evolution, which could be viewed as efficiency enhancing.
THE WAY FORWARD
While the fourth industrial revolution is already upon us, and algorithms and AI are omnipresent, competition law enforcers are still struggling to determine how to deal with this new reality. Disruptive technologies may have to adapt to competition law, but competition law will have to take into account new market realities, and enforcers will have to capture whether and where infringements are committed. Enforcement agencies may need to have recourse to AI themselves in detecting anticompetitive practices in future.
It is likely that AI and its applications will figure highly on the agenda of the new Commission when it takes office in November. All market players using AI will be watching what the policy agenda will bring, but the big data aggregators will likely have the most to fear.
Algorithms can create the “perfect price discrimination.”
Algorithms might entirely change the way suppliers interact with their customers in a market.