Hanna Gunn

In Neal Stephenson’s fictional novel, Diamond Age (1995), the protagonist Nell acquires a prototype of what we might today recognise as a highly sophisticated e-reader with a voice-assistant. This e-reader, the “Young Lady’s Illustrated Primer”, uses artificial intelli- gence to serve as Nell’s personal teacher. What is key to the Primer is how it is designed to respond to Nell. The Primer has a theory of Nell – her needs, her real-world situation, her abilities – and it tailors its lessons for her. The Primer is a highly personalised artificially intelligent device. For many readers – myself included – the Primer was a utopian vision of the bespoke digital tutors of the future. The appearance of new media and technologies in the early 2000s all pointed to the near-term reality of devices like Primers. The real-life experience of personalised digital media, though, is so far from Stephen- son’s vision that it is hard to imagine we will ever get there. Nell’s time with the Primer develops her intellectually, emotionally, and physically. Our “bespoke” media experience, by contrast, is more likely to turn us in on ourselves rather than guide us to live well in our societies. Personalised media is making us more like Narcissus: our consumption is driven by personal choices and automated algorithms that reflect what we already want to hear, see, and consume, and we are captivated. Central to our understanding of the problems with the personalisation of digital media are filter bubbles and echo chambers. These phenomena are used interchangeably by some, but in distinct ways by others. I take the position that filter bubbles and echo chambers are distinct phenomena, but ones that raise similar epistemic worries. I begin with a general discussion of filter bubbles and echo chambers, and I introduce the concept of “selective exposure” as central to both phenomena. There is minimal uniformity across uses of the terms “filter bubble” and “echo chamber”. Consequently, I divide the later discussion into two more precise topics: homophily and polarisation. I conclude with some future-looking comments about responsibility and the epistemic risks posed by digital personalisation.

ABSTRACT One of the promises of the internet was its power to unite individual knowers with one another, democratizing knowledge and spurring our collective efforts toward truth. In what sense is our current epistemic life a collective effort? This article examines the idea of the epistemic community. I contrast epistemic community with a collection of individual epistemic agents aiming for truth. I propose that this latter conception of epistemic life permits neglecting our epistemic and moral duties. I argue that healthy epistemic community requires joint commitments to truth and epistemic trust, and where epistemic trust requires conditions of mutual epistemic respect.

In this paper we take up the issue of near-term artificial intelligence (AI). "Near-term AI" is used to denote artificial intelligence algorithms that are already in place in a variety of public and private sectors, guiding decision that pertain to advertising, credit ratings, and sentencing in the justice system. Our focus here is to contribute to a critical discussion of the ways that AI is already being widely used in decision-making procedures in these areas. We will argue that developers and deployers of AI systems--in senses to be defined--bear a particular kind of responsibility for the moral consequences of near-term AI. We will present a tool to aid developers and deployers in engaging in the moral reflection we argue is required of them, in order both the help them meet their moral obligations and to help address the material risks posed by what we take to be the status quo of actual near-term AI development. This chapter can be understood as a contribution to the field of technology assessment, but instead of suggesting policy revisions, we will propose a framework for ethical analysis that can be used to facilitate more robust ethical reflection in AI development and implementation.

In a recent New Yorker cartoon, a man is fixing a sink. His partner, standing nearby skeptically asks, “Do you really know what you are doing, or do you only google-​know?” This cartoon perfectly captures the mixed relationship we have with googling, or knowing via digital interface, particularly via search engines. On the one hand, googling is now the dominant source of socially useful knowledge. The use of search engines for this purpose is almost completely integrated into many of our lives. On the other, the point the cartoon is making resonates with nearly all of us: users often recognize that there are risks and trade-​offs associated with gaining certain kinds of information via online search. These facts about googling make it particularly interesting to the applied epistemologist. Our practices involving search engines not only have a distinctive character, that character puts some traditional epistemic questions in a new light. This chapter will examine two of those questions. The first concerns the extent to which googling raises problems similar to familiar quandaries surrounding testimonial knowledge. The second –​and more radical –​concerns whether googling is a type of distributed or extended knowledge.

In a post-truth era, taking seriously the assertions of political figures and what other people say on the internet strikes many as irrational and gullible. Let us call this reaction the “incredulous reaction.” In this paper, I consider a common response to the targets of the incredulous reaction: excluding them from activities like debate and discounting their beliefs as relevant to our own. This exclusion is motivated by the assumption that those who continue to place epistemic trust in a post-truth society are acting irrationally. I begin with an argument that we can justifiably exclude irrational persons from our epistemic activities. I then present a conflict that emerges when we take the wellbeing of our epistemic community to be an end in and of itself, and not merely a means for pursuing other epistemic goods. I propose that healthy epistemic communities depend on networks of epistemic trust and on ensuring their members are treated with the epistemic respect they deserve. If we were to adopt the development of a healthy epistemic community as a goal, then excluding others may no longer be a justifiable choice.

If we’re aiming to have well-founded beliefs, then we generally think it’s a good idea to listen to a wide range of arguments. Listening to dissenting views, in particular, is important for avoiding epistemic dispositions to dogmatism and closed-mindedness. We might say we have an epistemic responsibility to listen to others in order to ensure we do our due diligence when forming beliefs. But do we also have to listen to the arguments of people with whom we disagree online? Received wisdom suggests no, and that not listening online is something of a life raft for epistemic self-preservation. Search engines like Google help to facilitate this self-preservation by filtering the testimony we will come across. There are legitimate concerns about the way that this sort of online personalization undermines our ability to access a sufficiently wide range of information. I explore here a related worry: how might our reliance on Google filtering be affecting our ability to listen responsibly and meet our duties to be good listeners? I propose that our received wisdom is misguided, and in fact search engines like Google have raised the stakes for what counts as good listening.

In this paper we explore whether or not the use of risk assessment tools in criminal sentencing can be made compatible with a retributivist justification of punishment. While there has been considerable discussion of the accuracy and fairness of these tools, such discussion assumes that one’s recidivism risk is relevant to the severity of punishment that one should receive. But this assumption only holds on certain accounts of punishment, and seems to conflict with retributivist justifications of punishment. Drawing on the broader desert literature, we explain the source of this conflict, and suggest that a retributivist approach on which the severity of punishment partly depends on one’s character in addition to their acts offers some hope of reconciling retributivism with the use of risk assessment tools in considerations of sentencing reduction in particular. Ultimately, however, even this limited attempt at reconciliation fails, so long as risk assessment tools fail to distinguish between risk that one is responsible for, and risk that one is not responsible for.

For most people, the internet is now the most dominant source of socially useful knowledge. Its widespread use has made knowledge more accessible, more widely distributed, and more commonly produced. But the internet is also widely seen—and not just by philosophers—as raising a number of distinct epistemological problems. Some of those problems concern the metaphysics of knowledge—the extent to which knowledge via the internet is understood as outsourced, or even extended, knowledge. Others concern the type of knowledge the internet can give us—whether, for example, the knowledge we gain by using our digital devices is a kind of testimonial knowledge. In this chapter, we will focus on a third issue: how our uses of the internet to gain information affect our epistemic agency—or our capacity to take responsibility for our own epistemically relevant mental states and our wider contributions to our epistemic environment. In the early days of digital technology, the internet was generally seen as increasing our epistemic agency for the simple reason that it made information (and presumably knowledge) more accessible and widespread. But in recent years, such optimism has been tempered due to the rise of fake news, massive amounts of misinformation online, and the average consumer’s seeming credulity with regard to what they read on social media. The chapter will outline these objections and critically examine them, arguing that both our epistemic agency and our ability to responsibly exercise such agency can be undermined by some uses of the internet, even as those same uses increase agency in other ways.

Finding ways to foster intellectual humility (IH)—the willingness to own one’s limitations—is an important goal for facilitating effective learning. We report the results of a longitudinal, quasi-experimental study, conducted across six undergraduate, culturally diverse (58% racial/ethnic minority) introductory philosophy courses, that evaluates how social perceptions and cross-cultural reasoning change following a course on epistemology and social ethics. Critically, we manipulated whether each class received a standardized lesson in IH at the start of the course or not. Participants provided self-ratings of IH, gave round-robin judgments of their classmates, and completed a wise-reasoning writing exercise focusing on a cultural conflict at both the beginning and end of the summer course session. Results revealed no change in self-reported IH, but an increase in perceived peer IH and conscientiousness within classes where IH was taught. Moreover, an analysis of the participants’ writing suggested greater levels of compromise-seeking when dealing with a cultural conflict among the students taught about IH, relative to those in the control classes. This study provides initial insights regarding the relationship between IH and cultural humility, as well as the efficacy of facilitating the development of IH among undergraduate students.