Executive summary
Rapid development of generative artificial intelligence chatbots like ChatGPT is leading search engine providers to move from search to answer engines. Unlike search engines, which provide search results in the form of blue links to content creators, answer engines generate personalised answers through a conversation with end users. This revolution impacts the internet ecosystem of content creators and the digital advertising market. This paper outlines some early antitrust issues related to answer engines, from the transition from search to answer engines (sections 2 and 3) and the response competition authorities should adopt (section 4). It finds that search and answer engines complement and compete with each other. While the answer-engine market is still at an early stage of development, it already raises some competition issues in relation to data scraping, vertical integration and unfair terms and conditions. Intervention by competition authorities is more likely than not to prevent market power in this new market. In this regard, competition authorities should act to preserve dynamic competition and minimise adverse effects on content creators. Finally, the paper concludes with several research questions for future research (section 5).
Introduction
The ultimate search engine, “would understand everything that you asked it, and give you back the exact right thing instantly,” said Larry Page, the co-founder of Google, back in 2006. Until now, search engines have provided what end users want in the form of organic search results of lists of content creators via blue links. But the advancement in generative artificial intelligence (AI) chatbots forces search engine providers to move from being search engines to being answer engines. Unlike search engines, generative AI answer engines generate an answer through a conversation with the end user (a so-called ‘conversational answer’). Answer engines drastically diminish search costs and increase productivity for end users because they provide what users want directly without the need to click on search results. This revolutionary change from search to answer engines impacts the internet ecosystem of content creators and the digital advertising market.
Internet actors, including search engines and content creators, rely on blue links to get traffic and generate revenue. Search engine providers rank blue links and derive revenue from advertising by promoting sponsored blue links above organic search results – a market worth $260 billion in 2022 based on worldwide search advertising spending. Some search engine providers also derive revenue from non-search advertising, including banner or video ads, by offering advertising services, such as ad inventory, to content creators (CMA, 2020). Google thus generated about $224.5 billion in 2022 from digital advertising, including $162.5 billion from search advertising and $62 billion dollars from nonsearch advertising. Content creators also generate revenue from digital advertising thanks, in part, to the traffic they receive from search engines. If answer engines lead to significant traffic loss to blue links, search engines and content creators will lose advertising revenue.
Nevertheless, despite the revenue loss, search engine providers cannot avoid moving from search to answer engines. They cannot afford not to compete because newcomers, such as ChatGPT, threaten ‘traditional’ search engines, forcing the latter to innovate. ChatGPT, provided by Open AI, is the leading answer engine. In January 2023, after only two months of existence, ChatGPT already reached 100 million monthly active end users. As a comparison, it took 2.5 years for Meta-owned Instagram to gain 100 million users. The move to answer engine is thus a matter of survival for search engines. Not competing means a risk of being displaced by new entrants in a new market.
In this context, search and answer engines are both complementary and competing. They complement each other as they support each other to run. Search engines need answer engines to survive. Conversely, answer engines need search engines to provide references to the generated answer.
They compete as they both provide what users want. It follows an intense dynamic competition for the transition from search to answer engines by incumbent search engine providers and new entrants. Nonetheless, answer engines already pose several competition issues. Indeed, answer engines collect and use data, including copyrighted data, to train their models and generate revenue. In jargon, they scrape the web, sometimes without permission, and use data as input from content creators to generate valuable content, in direct competition with content creators. They thus raise the old but still unsolved competition issue of data scraping. Answer engine providers, such as Microsoft, also integrate their answer engines into their own search engines. They promote their answer engines at the top of search results, and thus above the content of rivals. This raises the competition issue of self-preferencing when a firm promotes its own services over rivals. Some developed economies, including the European Union, have opted for a ban on self-preferencing in some circumstances. The ban will likely impact the development of answer engines. Finally, some answer engines like Microsoft rely on search-engine technology. In particular, they combine index data and generative AI technology to provide an up-to-date answer with citations. However, Microsoft is also one of the two providers of index data in the world. According to press reports, Microsoft has threatened to cut access to its index data for rivals that use its index data to develop their own answer engines. If implemented, the threat would be akin to a refusal to supply input to rivals with unfair terms and conditions, thus raising the complex competition issue of refusal to supply.
The answer engine market is still nascent, with frequent innovations and market entries. At this stage, the competitive process works well. Despite some competition issues, there are not yet market power issues that would require competition authorities to intervene. Nevertheless, intervention will more likely than not be needed to prevent market power in this new market. Competition authorities should preserve the intense competition in the transition from search to answer engines. They should monitor closely that incumbent search engine providers do not impose entry barriers on new entrants in the answer engine market, in order to protect their business. They should also study the impact of answer engines on the internet ecosystem and how end users behave in relation to content creators. In particular, they should conduct behavioural economic studies on how users interact with blue links, because these are content creators’ primary source of traffic and revenue. In case of adverse effects on content creators, competition authorities should monitor the compensation mechanisms offered by answer engines that send traffic back to content creators. If these compensation mechanisms are ineffective, they should consider working with stakeholders to develop effective alternatives.
The paper thus outlines some early antitrust issues related to the transition from search to answer engines (sections 2 and 3), and the response that competition authorities should adopt (section 4). While the answer engine market is still nascent and unpredictable, it raises fundamental research questions (section 5). Competition experts should study in depth the impact of answer engines on the internet ecosystem. They should also explore how large online platforms are integrating generative AI into their ecosystems, in particular in cloud and software services, and any attempts to lock-in their end users and business users into their ecosystems. Finally, competition authorities should pay attention to how the advancement of generative AI technology and legal regimes, including competition, data protection, content moderation and AI, impacts the evolution of answer engines.
About the Author
Dr. Christophe Carugati is an affiliate fellow at Bruegel on digital and competition issues.