Book Review: Regulating the Synthetic Society: Generative AI, Legal Questions, and Societal Challenges

This is a review of “Regulating the Synthetic Society: Generative AI, Legal Questions, and Societal Challenges” by Bart van der Sloot (Tilburg University). 

The book is divided roughly into distinct parts: the first three chapters go into some detail to describe various AI technologies, how they work and their real-life applications (both present and future) whereas the latter three chapters set out the challenges associated with these technologies, how current legislation seeks to regulate them (and where it fails to do so) and some proposals for “imperfect solutions” to address “uneasy questions”. In more detail: 

In Chapter 1, “A Map to this Book”, van der Sloot clarifies that the book focuses on four synthetic technologies, namely “Deepfakes” (content that has been wholly or partially fabricated, or manipulated, using AI), “Humanoid Robots” (robots with human appearance and human-like cognitive functions), “Augmented Reality” (reality as perceived by the user of a particular piece of equipment that adds layers to reality using AI) and “Virtual Reality” (where all sensory input is AI-generated). 

Van der Sloot notes that, given that much of the information we have on AI comes from computer scientists or tech-savvy journalists/academics, these sources tend to be biased towards benevolent use cases and positive impact projections. 

In Chapter 2, “Applications and Effects of Synthetic Technologies”, van der Sloot sets out many examples of the current and potential use cases for the above technologies, roughly categorised into five domains: 

• Medical: e.g. Augmented Reality (AR) assisting surgeons in carrying out precision tasks in surgical procedures, Humanoid Robots (HR) being assigned non-patient-related tasks such as disinfection;
• Industry/commerce: e.g. AR being used by workers in assembly lines to receive additional information (as first popularised by Boeing), and Deepfakes (DFs) facilitating online cross-border conversations by translating in real time into each person’s language;  
• Education: e.g DFs being used to deliver interactive lessons to students (Napoleon speaking about the French revolution), or VR bringing schoolchildren back to ancient sites; 
• Entertainment: e.g. VR being used for gaming; and
• Security: e.g. DFs being used by the police to virtually infiltrate criminal networks, and AR being used for military training. 

Chapter 3, “Under the Hood: Architecture and Design of Synthetic Technologies”, provides a high level overview of the technical design and architecture of these technologies, explaining, for example, that the objective in respect of HRs is for them to be indistinguishable from human intelligence and appearance, and for VR to replace physical reality in full (including through the further development of haptic input and output systems and so-called ‘soft robotics’ such as AI scent generators to stimulate olfactory and sensory inputs).

Having set out some of the use cases for synthetic technologies, Chapter 4, “Societal Challenges”, sets out some of the issues which are likely to arise, focusing primarily on the long-term effects and societal implications. 

The most significant risks arise out of AI’s potential to be an additional layer between people and reality, by making a virtual reality much more plausible and convincing. For example, more realistic DFs spreading fake news may erode society’s trust in the media and the objective concept of truth. Virtual contact between human beings may entirely replace physical / in-person contact, with significant effects on human development, and humans may become increasingly dependent on technology and consequently less autonomous. 

In Chapter 5, “Law”, the book sets out some of the legal regimes (with a focus on the EU) which may have an impact upon the various synthetic technologies set out above, and it rapidly becomes clear that the potential scope is vast. Only in the EU, laws and regulations which may be of relevance to AI include the GDPR, the AI Act, the AI Liability Directive, Product Liability Directive, Digital Services Act, Dual Use Regulation, the Machine Regulation and more! 

One particularly salient question raised by van der Sloot is whether the current regulatory trend of ex post regulation (including, for example, by requiring tech companies to self-regulate) remains appropriate in respect of such fast-moving and all-pervasive technologies. 

Finally, in Chapter 6, “Uneasy Questions and Imperfect Solutions”, van der Sloot seeks to offer some options for regulators going forwards. 

This book is an extremely interesting read. Although one can hardly get through a day without seeing some mention of AI, much of the commentary in the general press tends to revolve around LLMs and their potential impact within the professional services environment, so it is easy to forget that AI technologies can manifest themselves in many different guises, and will have the ability to impact (for the most part, positively) a large variety of sectors, including medicine and education. 

Whereas van der Sloot necessarily engages in some speculation, given the relative infancy of some of the technologies in question and the pace of development, the book provides an invaluable base for anyone involved or interested in regulation, to begin understanding the issues which regulators are likely to have to consider in future (particularly in the medium-term).  

Available in Hardback & Paperback

Extent 296 pages

ISBN 978-1-50997-494-8 

Publisher Bloomsbury Publishing