In the last few decades, the growing and crosscutting influence of digital technologies has had an impact on many existing research areas and has led to the flourishing of new research areas too, such as the ethics of artificial intelligence (AI) and roboethics. Most of the related research efforts in these areas are devoted to analyzing the ethical dimension of information and communication technologies (Floridi 2014), and of AI and algorithms (Liao 2020; Mittelstadt et al. 2016; Tsamados et al. 2021), and to devising morals applicable to intelligent machines (Floridi and Sanders 2004; Bostrom, forthcoming).

In this general framework, growing attention has been paid to the ethical design of digital technologies, where AI systems are considered as both objects (i.e., tools for/made by humans) and potential subjects (i.e., moral agents and patients): proposals aimed at developing ethics by design or value-sensitive design approaches to digital technologies (van den Hoven, Vermaas and van de Poel 2015; Friedman, Hendry and Borning 2017; Umbrello 2020) have significantly increased in the last years, especially in the framework of the design of AI for the social good (Umbrello and van de Poel 2021).

At the same time, attention has increasingly been paid also to specific ethical issues linked to digital technologies, such as privacy (Nissenbaum 2004; Richards 2015); tracking, monitoring, and processing of users’ data (Wolmarans and Voorhove 2022); the impact on personal autonomy and identity (Botes 2022); opacity (Bonicalzi 2022; Herlocker, Konstan and Riedl 2000); biases and unfairness (Buolamwini and Gebru 2018; Eubanks 2018; Noble 2018; O’Neil 2016; Zimmermann and C. Lee-Stronach, 2022); manipulation (Klenk and Hancock 2019); the potential threats to democracy and society as a whole (Christiano 2022; Risse 2023); and the prospective loss of jobs and unemployment (Ernst 2022), just to mention the most relevant ones.

A key underexplored question concerns how and to what extent digital technologies may foster or hinder individual and collective human decision-making. This is a crucial issue to the extent that digital technologies shape the reality we live in, and affect the pre-conditions of our (supposedly free) choices: the progress and use of machine learning algorithms, based on deep learning architectures, often lead to non-explainable algorithmic decision-making (Pasquale 2015), involve biases (Benjamin 2019), and tend to create filter bubbles (Pariser 2011) or echo chambers (Sunstein 2008), thereby affecting our epistemic agency (Coeckelbergh 2022), predetermining the conditions and restricting the range of our choices (Giovanola and Tiribelli 2022). Moreover, we are delegating a great deal of the choice process to digital technologies, often without even realizing it (Royakkers et al. 2018) but nonetheless contributing to the gradual erosion of our moral capacities.

This situation raises underexplored ethical concerns and questions about the possibilities and constraints of human decision-making, in a reality that is more and more shaped by artificial intelligence and algorithms in a wide array of application domains, including—among others—social media communication and information management (Bozdag 2013; Shapiro 2020; Hinman 2008), advertising and marketing (Hildebrandt 2008; Tufekci 2015), recruiting and employment (Kim 2017), university admissions (Simonite 2020), housing (Barocas and Selbst 2016), credit lending (Devill 2013; Lobosco 2013; Lee and Floridi 2020), criminal justice (Berk et al. 2018), policing (Ferguson 2017), and healthcare (Danks and London 2017; Robbins 2019; Giovanola and Tiribelli 2023; Migliorelli et al. 2023).

To sum up, digital technologies increasingly shape the reality we live in as well as the boundaries of our moral agency, and raise the question of what the properly “human” side of decision-making is. At the same time, digital technologies have a pervasive impact on many dimensions of decision-making, which are increasingly automatized, supported by or delegated to AI and algorithms. Both aspects have been intensively investigated in isolation. However, to date little effort has been made to explore their connections—i.e., to investigate how digital technologies foster or hinder decision-making, and to examine whether AI and algorithms affect the pre-conditions of our choices and moral agency.

This edited collection aims to contribute to filling these research gaps, bringing together philosophers and AI experts to analyze the interplay between digital technologies and human decision-making.

The articles included in the present collection offer an analysis of the existing and foreseeable ethical and social challenges that AI technologies posit to human decision-making. Several contributions focus on concerns linked with specific emerging digital tools and areas of application—including online nudging (Schmauder, Karpus, Moll, Bahrami and Deroy), mental healthcare (Minerva and Giubilini), and recommender systems (Bonicalzi, De Caro and Giovanola)–, or with their techno-social implications, such as manipulation (Ienca) and increase in datafication (Lavazza and Farina). Other papers aim at critically overviewing or developing ethical frameworks and guidelines (Liao; Mhlambi and Tiribelli; Deroy), as well as political responses (Hoeksema), to be deployed in practical contexts where the usage of AI-based technologies is, or will be, skyrocketing.

Christian Schmauder, Jurgis Karpus, Maximilian Moll, Bahador Bahrami and Ophelia Deroy explain how AI methods provide new avenues, e.g., through an increase in fine-tuned personalization, for the development of effective forms of nudging. Nudge is a public policy that may often facilitate or improve human life across different decision-making domains, spanning from healthcare to the protection of the environment. However, in their contribution, the authors carefully highlight that outsourcing the design of nudges to AI systems whose workings are not entirely known or explainable—the infamous black box problem associated with AI—might be socially undesirable or even pernicious, with associated problems of accountability. In particular, since nudges notoriously exploit human weaknesses and fallacies, the obscure nature of AI-based nudge techniques may imply that users and programmers be unaware of the human cognitive processes that are selectively involved in reaching specific decision-making targets. On these grounds, the authors advocate for an interdisciplinary monitoring of the AI systems designing nudges.

In their contribution, Francesca Minerva and Alberto Giubilini consider whether AI is to be thought of as the future of mental healthcare. On the one hand, the possibility of deploying AI-based tools has considerably helped the administration and accessibility of healthcare worldwide. This appears to be particularly important in an age when mental health is globally deteriorating, demands for help are progressively augmenting, and AI might be able to outperform, or at least proficiently assist, human doctors throughout several medical procedures. Nonetheless, switching to forms of AI-based healthcare has some significant drawbacks, including, for instance, the dehumanization of the medical procedures. Thus, Minerva and Giubilini carefully address the pros of cons of using AI in the delivery of mental healthcare.

In their paper, Sofia Bonicalzi, Mario De Caro and Benedetta Giovanola contribute to the philosophical discussion concerning the ethical issues that have been raised in relation to the widespread diffusion of recommender systems throughout various life activities. In particular, they target the impact that recommender systems may have for what they term descriptive autonomy, a multi-faceted notion including both the potential to express oneself through action and the capacity to exert reasons-responsiveness and engage in reflective practices. Advocating for an ethically oriented implementation of recommender systems, but without indulging in the unrealistic defense of the status quo, the authors further articulate this challenge in terms of the risks of manipulation and deception associated with recommender systems, of their power to affect users’ personal identity, and of their impact on knowledge acquisition and sharing processes as well as on critical thinking.

The focus on users’ manipulation as a byproduct of the new forms of engagement brought about by the development of AI systems is central also to Marcello Ienca’s contribution. Here, the author provides an insightful critical overview of the literature on manipulation and AI technologies. Throughout the piece, Ienca highlights how manipulation is not uniquely associated with AI and its cognate tools. Indeed, AI-mediated forms of manipulation are not to be represented as qualitatively different from analogous dynamics occurring in human–human interactions. At the same time, they have unprecedented potential in terms of their capacity to target and steer people’s decision-making, through their aptness to bypass users’ cognitive defenses. On these grounds, the author discusses how various social actors, such as researchers, practitioners, and policymakers could deal with such challenges.

In our increasingly digitalized society, recommender systems (Bonicalzi, De Caro and Giovanola) and online nudging (Schmauder, Karpus, Moll, Bahrami and Deroy) represent just one of the many technological artifacts and tools with which users are learning to interact, and that are reshaping, to a large extent, their experience across different social environments. In their contribution, Andrea Lavazza and Mirko Farina provide a comprehensive overview of the ethical and social qualms associated with the intensive datafication that underlies the implementation of intelligent and sophisticated bio-technological unions. Their extensive analysis is complemented by a reflection on the desirability of the profound changes that such datafication processes will bring about. The discussion takes as a basis four fundamental insights regarding the impact that AI-based technologies may have on the experience of users and, more generally, citizens and workers. These respectively concern the tendency to erode human privacy, which may expand into forms of worrisome social and political control, the reduction of workers’ freedoms, the limitations imposed on human creativity and imagination, and the overemphasis on efficiency and instrumental reason.

AI-based technologies deployed in sensitive social contexts, such as healthcare (see also Minerva and Giubilini), are under pressure in terms of the development of appropriate ethical frameworks and guidelines. In recent years, various social actors, ranging from private subjects to governmental agencies and institutions, have contributed to producing such documents. In reviewing them, S. Matthew Liao aims to go beyond the existing formalizations, emphasizing their limitations and flaws. While these normative tools share some fundamental principles, including the focus on autonomy and non-maleficence, they tend to be too abstract to have a profound impact on digitally advanced healthcare sectors and address the multiple concerns raised by AI-based technologies. Furthermore, at a more theoretical level, they often do not manage to appropriately justify the very same principles that they defend. To bridge these practical and theoretical gaps, Liao proposes an ethical framework that finds its proper justification in human rights theory, which is aptly extended to the healthcare domain and holds that people have rights to “the fundamental conditions for pursuing a good life”.

The focus on the limitations of the current ethical frameworks for AI is central to Sábëlo Mhlambi and Simona Tiribelli’s contribution as well. Referring to the existing ethical frameworks, the authors provocatively call into question the almost exclusive emphasis on a liberal notion of autonomy as self-determination (see also Bonicalzi, De Caro and Giovanola). This notion is criticized as inadequate to account for the many senses in which human autonomy can be violated in the context of artificial decision-making. Furthermore, as this narrow notion of autonomy is grounded in Western traditional philosophy and linked with a history of colonization, the corresponding ethical frameworks may fail to understand the extent to which AI-related harms may cause substantial trouble to those who are already globally marginalized and disenfranchised. In the attempt to respond to this cultural challenge, Mhlambi and Tiribelli stress the need for a relational turn, rooted in moral philosophy and Ubuntu ethics, in the ethical frameworks regulating AI.

Another controversial notion associated with AI technologies and consistently deployed by ethicists, governmental agencies, and institutions is the label “trustworthy AI” In her contribution, Ophelia Deroy warns against the indiscriminate usage of ambiguous or controversial expressions, or loose talk, attributing human-like features to AI. By browsing the field of AI ethics and science communication, the author reviews the reasons and speaks out against the fragile justifications—ontological, legal, communicative, and psychological—that contribute explaining these questionable linguistic practices. In particular, in pointing at the potentially negative social consequences that this attitude may have, Deroy focuses on two problematic arguments that may underlie the tendency to anthropomorphize AI, i.e., the claim that discourses on the ethics of AI do not fundamentally require philosophical clarification, and the claim that such humanizing language appropriately matches how non-experts conceptualize AI.

While most of the papers hosted in this collection are concerned with the ethical challenges posited by digital technologies, Bernd Hoeksema’s contribution aims to articulate a political perspective, under the umbrella of republicanism, on online jerkish speech. Jerkish speech is here identified as the speech with which users show disregard for the perspective of others, notably when the latter are perceived as having a lower social status. While online jerkish speech might be wrongly considered scarcely impactful in comparison with forms of more explicit hate speech or online harms, it could lead to systemic or structural forms of domination or forms of micro-domination, the latter being individually inconsequential but nonetheless problematic when they are thought of in aggregate. In the paper, Hoeksema discusses how the republicanism program, having a focus on the notion of “domination”, has the tools to explain why online jerkish speech is problematic as well as to develop an appropriate strategy to reduce its social impact.

Thus, as this brief overview aims to illustrate, this collection on Digital Technologies and Human Decision-Making aims to raise a discussion about the most urgent ethical and social challenges regarding the impact digital technologies (may) have on human decision-making, both for individuals and for society as a whole.