Digital Platforms and Machine Yearning:Longing for “Big Tech” to Deliver on AI's Promise

JP Vergne – UCL School of Management

-- Make no mistake, the title of this post does not contain a typo.

For the past 15 years, the corporations at the forefront of AI development, such as Google, Facebook, and Amazon, have made promises that they cannot keep. To their users, they promised universal digital platform access to information, people, and goods (respectively), no matter where the latter are initially produced or located around the globe. To their customers, namely advertisers and merchants, they promised access to a global pool of consumers whose preferences and purchasing behavior can be predicted using AI – more specifically, using machine learning (ML) algorithms trained on huge sets of personal data collected from platform users.

Instead of top-down, hierarchical relationships between people and businesses, we were promised horizontality and decentralization. Your online browsing activity makes Google’s search engine better and so does everyone else’s, on equal footing. By the same token, anybody can advertise something on Facebook to someone else – no need to be a corporation or to hire an advertising agency. And any merchant can register with amazon.com to reach consumers anywhere in the world. And so on.

Yet, as I explain in “Decentralized vs. Distributed Organization: Blockchain, Machine Learning and the Future of the Digital Platform,” recently published in Organization Theory, underneath the appearance of horizontality and decentralization lurks a different reality. To operate at scale a business whose revenues depend on the sale of ML predictions, corporations must aggregate, concentrate, and centralize both data and their processing. This is affecting every industry, including professions such as medicine and the law, as argued recently by David C. Donald.

The gaping discrepancy between the false promise of data decentralization and the carefully concealed reality of data gravity has shaped the contours of a dystopian digital society. We could all become TikTok stars, yet the probability of that happening to a given individual is only known to ByteDance, the corporation that owns the platform and centrally processes all user data on its private servers. All these years, we have been yearning for AI-driven decentralization in vain because we have conflated decentralization with something else – let’s call it “distribution.”

While decisions to search for particular keywords on google.com are distributed across billions of users, the collection and processing of valuable data resulting from search decisions must remain centralized for subsequent AI processing. In other words, the digital platforms operated by “big tech” are not decentralized, but merely distributed. Their centralization keeps increasing, in fact, as I argue in the above-mentioned article in Organization Theory.

An alternative technological blueprint has emerged over the past decade to enable the design of truly decentralized platforms (which are also distributed). It is called “blockchain” and, as strange as it may seem, it does represent an alternative to ML technology for creating digital platforms. Bitcoin and Ethereum are the two most prominent examples of such decentralized platforms.

Regulatory implications cannot be understated. Just like clever traffic regulation takes into account the existence of both cars and bicycles, clever regulation of the platform economy needs to balance competition, innovation, and privacy issues within and between centralized and decentralized platform ecosystems.

To do that, we must deepen our understanding of decentralization so we can better measure it. Until now, our insufficient understanding of the notion has led us to believe that breaking up, rather arbitrarily, a couple of trillion-dollar corporations could solve our problems. It cannot and it won’t.

It’s time to stop yearning for a smaller Google or a smaller Amazon and to start learning what’s different about AI and blockchain that makes our prevailing view on antitrust, inspired by a U.S. law written two centuries ago, obsolete and deeply misleading.

107 views0 comments

Recent Posts

See All

Judging Autonomous Vehicles

Jeffrey J. Rachlinski – Cornell Law School; Andrew J. Wistrich – California Central District Court -- Would you rather be run over by a self-driving car or a car driven by a human being? Assuming a s

What is Legal Innovation?

Haim Sandberg – School of Law, The College of Management Academic Studies -- Is law itself an arena of innovation – of legal innovation? Can we learn something about the nature of legal innovation fr

Measuring Law Over Time

Corinna Coupette, Max Planck Institute for Informatics; Janis Beckedorf, Ruprecht-Karls-Universität Heidelberg; Dirk Hartung, Center for Legal Technology and Data Science; Michael Bommarito, CodeX – T