What should governance reconsider as digital technologies change?
Rules for new technologies are usually made before anyone knows how those technologies will play out. But what should happen once we learn more? This question is at the heart of a new series of essays by Urs Gasser, who has joined Science’s “Expert Voices”.
In several essays a year, Urs Gasser will explore how institutions make decisions amid technological uncertainty – and how they reconsider those decisions as knowledge and circumstances change. Drawing on examples from AI, law, and public policy, the pieces explore what it takes to make learning an integral part of how we govern.
His first essay starts with a question many will recognize: Can you ask an AI to forget you? Since a landmark 2014 ruling, people in Europe can, under certain conditions, have search engines remove links to personal information about them – the so-called right to be forgotten. But generative AI systems do not simply return lists of links, and removing a link does not map neatly onto a model that generates rather than points. Does the law need to change again? Perhaps, Gasser writes – but the question bundles together different things that might, or might not, need changing.
The essay distinguishes between three elements of a governance arrangement that may need reconsideration as digital technologies change: its underlying purpose or value, the legal rule through which that purpose is expressed, and the operational mechanisms through which the rule is implemented. Gasser refers to the diagnostic process of determining which of these elements, if any, no longer fits changing circumstances as “revalidation”. New knowledge does not, by itself, indicate what should be changed. Rather, it provides reasons to examine whether the purpose, the rule, or its implementation remains appropriate. Importantly, revalidation does not necessarily lead to change. It may result in modification at one level while leaving another level unchanged, or it may confirm that the existing arrangement remains appropriate. In this sense, stability following deliberate reconsideration differs from stability by default.
Gasser also highlights a question of power: the actors who possess the knowledge and authority needed to revalidate an arrangement may be the same actors whose interests, resources, or business models depend on its continued operation. Revalidation is therefore not only a matter of adapting governance to technological change, but also of determining who has the authority to decide what should be reconsidered and on what basis.
This essay is the first in a series. Science launched “Expert Voices” for topics that need more space than its Editorial page allows. In Science’s words, the feature brings together “a stable of columnists who will write several times a year on matters of concern within their fields that are also likely to interest nonexperts.” Urs Gasser’s contributions will focus on how institutions make decisions amid technological uncertainty – and how they reconsider those decisions as knowledge and circumstances change, drawing on examples from AI, law, and public policy.
The essay acknowledges funding from the DFG Cluster of Excellence TransforM and the MacArthur Foundation.