Abstract
Whenever we go online, we are bombarded by advertisements whose placement is informed by our browser history. Are we being ‘controlled’ by the algorithm which led to those ads? What about algorithms that preference news content into our feed which reinforces our algorithmically predicted predispositions, creating strident echo chambers and a belief that other sources are ‘fake news’?
Scherz rightly draws our attention to what might be called the ‘algorithmic state’ – where more of our life is influenced by predictive models, driven by data about our – and our cohort’s – behaviour. The risk is that predictive analytics and algorithms lead to dehumanization – where lives are controlled by the predictive algorithm, and where human agency is diminished. Scherz makes the bold claim that ‘probability becomes an attempt to erase the individual, to erase the particular free subject and submerge him in a predictable population’ (p. 90).
Scherz’s book appears at the same time as there is increasing attention to the power of large language models, such as ChatGPT, which can emulate human behaviour, including passing medical and legal exams. Will these models be an adjunct to improve decision making by doctors and lawyers, or will they replace human agency, or a bit of both?
Tomorrow’s Troubles is structured in three parts. The first canvasses aspects of risk – the history of its use, some basic concepts and some implications of risk analysis. The second part builds on this to highlight how risk prediction is being used today.
I have some quibbles here, principally about the hyperbole in the book, including whether we are controlled by these applications, which is the premise of Scherz’s argument, or whether the use of predictive models simply informs. Certainly, control is possible, and Scherz cites an example where a human agent tragically refused to override algorithmic advice, but is this the norm or a terrible aberration? Will that be the use in the future? Scherz certainly hopes not, and in the third part of the book he posits an approach that potentially will reduce adverse impacts.
Scherz, an associate professor in moral theology at the Catholic University of America, inserts virtue ethics as a frame for thinking about the use case of ‘algorithmic governance’. He highlights how a Christian understanding might lead to a different emphasis from the prioritization of the automaton and the algorithm. A better frame, he argues, would emphasize human agency, the relational nature of and priority of humanity and love, and creation. Scherz aims to rebalance the risk and algorithmic society back to a human- and God-centred frame, emphasizing prudence rather than calculus. In so doing he provides a useful antidote to the triumphalism of AI as it is currently portrayed in popular literature. However, the many benefits of predictive analytics and judgements are de-emphasized in his analysis and he goes too far in, for example, rejecting cost–benefit analysis (p. 211) without discussing what might be its place, subject to appropriate controls. Overall, though, this is a book that makes a useful contribution to the ethics of the increasing use of predictive analytics and algorithms in contemporary society.