-
Expand entry
-
Added by: Clotilde Torregrossa, Contributed by: Marilou NieddaAbstract:
This paper argues that AI follows classical versions of epistemology in assuming that the identity of the knowing subject is not important. In other words this serves to ‘delete the subject’. This disguises an implicit hierarchy of knowers involved in the representation of knowledge in AI which privileges the perspective of those who design and build the systems over alternative perspectives. The privileged position reflects Western, professional masculinity. Alternative perspectives, denied a voice, belong to less powerful groups including women. Feminist epistemology can be used to approach this from new directions, in particular, to show how women’s knowledge may be left out of consideration by AI’s focus on masculine subjects. The paper uncovers the tacitly assumed Western professional male subjects in two flagship AI systems, Cyc and Soar.
Comment: This text is very useful to offer a more socially situated account of AI to students: it is fairly easy to comprehend and it makes a very important point about the fact that symbolic language is not socially neutral.
-
Expand entry
-
Abstract:
The recent proliferation of deepfakes, AI and other digitally produced deceptive representations has revived the debate on the epistemic robustness of photography and other mechanically produced images. Authors such as Rini (2020) and Fallis (2021) claim that the proliferation of deepfakes pose a serious threat to the reliability and the epistemic value of photographs and videos. In particular, Fallis adopts a Skyrmsian account of how signals carry information (Skyrms, 2010) to argue that the existence of deepfakes significantly reduces the information that images carry about the world, which undermines their reliability as a source of evidence. In this paper, we focus on Fallis’ version of the challenge, but our results can be generalized to address similar pessimistic views such as Rini’s. More generally, we offer an account of the epistemic robustness of photography and videos that allows us to understand these systems of representation as continuous with other means of information transmission we find in nature. This account will then give us the necessary tools to put Fallis’ claims into perspective: using a richer approach to animal signaling based on the signaling model of communication (Maynard-Smith and Harper, 2003), we will claim that, while it might be true that deepfake technology increases the probability of obtaining false positives, the dimension of the epistemic threat involved might still be negligible.
Comment: It'd be a good reading for a class that touches on the epistemic challenges of AI and digital images. What is fun and interesting about it is that (i) it gives a more positive view on the issue: it argues that the epistemic threat of AI and manipulated images is not as terrible as many suggest and (ii) in order to argue for this position it relies on the animal world.
-
Expand entry
-
Added by: Alnica VisserAbstract:
John Searle, in his paper on 'Minds, Brains, and Programs' (1980), argues that computational theories in psychology are essentially worthless. He makes two main claims: that computational theories, being purely formal in nature, cannot possibly help us to understand mental processes; and that computer hardware-unlike neuroprotein-obviously lacks the right causal powers to generate mental processes. I shall argue that both these claims are mistaken.
Comment: Excellent summary of the Chinese Room argument along with some interesting objections. Can be used as a follow-up reading to Searle, but also in place of it.
-
Expand entry
-
, Contributed by: Marilou NieddaAbstract:
In this chapter, I draw on the seminal work of feminist theorist Iris Marion Young to consider how political responses to structural injustice, exacerbated by AI,¹ might be developed. My interpretation of Young’s work is centred on what I see as an overlooked feature of her account of structural injustice - ‘untraceability’. Drawing on Arendt’s concept of ‘thoughtlessness’ and the work of other scholars such as Benjamin, I explore the relationship between structural injustice and algorithmic decision making,² which, itself, is becoming increasingly untraceable. I argue that while fault-based regulation of AI is vital for combating injustice exacerbated by AI, too little thought goes into addressing the structural dynamics of AI’s impact on society which requires a different sort of approach. I conclude by offering some suggestions on how we might think about mitigating the exacerbation of structural injustice posed by AI. In particular, I advocate adopting several elements of the mini-public approach within the regulatory public-body landscape of democratic governance systems to form a new ‘AI Public Body’ (using the UK as my example).Comment: On the one hand, this book chapter is useful for introducing the concept of structural injustice to students by examining the (un)traceable responsibility that individuals bear when they unconsciously (re)produce structural inequities. The text is fairly easy to read, especially to apprehend such a central concept in feminist theory of justice. On the other hand, Browne applies her understanding of traceable liability to AI technologies, which is an application that students can understand quite well. It also brings about a social perspective in the philosophy of AI.
-
Expand entry
-
Added by: Fenner Stanley TanswellAbstract:
During the 1970s and 1980s, a team of Automated Theorem Proving researchers at the Argonne National Laboratory near Chicago developed the Automated Reasoning Assistant, or AURA, to assist human users in the search for mathematical proofs. The resulting hybrid humans+AURA system developed the capacity to make novel contributions to pure mathematics by very untraditional means. This essay traces how these unconventional contributions were made and made possible through negotiations between the humans and the AURA at Argonne and the transformation in mathematical intuition they produced. At play in these negotiations were experimental practices, nonhumans, and nonmathematical modes of knowing. This story invites an earnest engagement between historians of mathematics and scholars in the history of science and science studies interested in experimental practice, material culture, and the roles of nonhumans in knowledge making.Comment (from this Blueprint): Dick traces the history of the AURA automated reasoning assistant in the 1970s and 80s, arguing that the introduction of the computer system led to novel contributions to mathematics by unconventional means. Dick’s emphasis is on the AURA system as changing the material culture of mathematics, and thereby leading to collaboration and even negotiations between the mathematicians and the computer system.
-
Expand entry
-
Added by: Alnica VisserAbstract:
Since its beginning in the 1950s, the field of artificial intelligence has cycled several times between periods of optimistic predictions and massive investment (“AI spring”) and periods of disappointment, loss of confidence, and reduced funding (“AI winter”). Even with today’s seemingly fast pace of AI breakthroughs, the development of long-promised technologies such as self-driving cars, housekeeping robots, and conversational companions has turned out to be much harder than many people expected. One reason for these repeating cycles is our limited understanding of the nature and complexity of intelligence itself. In this paper I describe four fallacies in common assumptions made by AI researchers, which can lead to overconfident predictions about the field. I conclude by discussing the open questions spurred by these fallacies, including the age-old challenge of imbuing machines with humanlike common sense.
Comment: Short easy read. Pairs well with Turing, giving a good summary of the technological progress that has been made since the 50s along with a more pessimistic interpretation of the theoretical import of the progress.
-
Expand entry
-
Added by: Chris Blake-Turner, Contributed by: Thomas HodgsonAbstract: One of the most important aspects of Grice's theory of conversation is the drawing of a borderline between what is said and what is implicated. Grice's views concerning this borderline have been strongly and influentially criticised by relevance theorists. In particular, it has become increasingly widely accepted that Grice's notion of what is said is too limited, and that pragmatics has a far larger role to play in determining what is said than Grice would have allowed. (See for example Bezuidenhuit 1996; Blakemore 1987; Carston 1991; Recanati 1991, 1993, 2001; Sperber and Wilson 1986; Wilson and Sperber 1981.) In this paper, I argue that the rejection of Grice has moved too swiftly, as a key line of objection which has led to this rejection is flawed. The flaw, we will see, is that relevance theorists rely on a misunderstanding of Grice's project in his theory of conversation. I am not arguing that Grice's versions of saying and implicating are right in all details, but simply that certain widespread reasons for rejecting his theory are based on misconceptions.1Relevance theorists, I will suggest, systematically misunderstand Grice by taking him to be engaged in the same project that they are: making sense of the psychological processes by which we interpret utterances. Notions involved with this project will need to be ones that are relevant to the psychology of utterance interpretation. Thus, it is only reasonable that relevance theorists will require that what is said and what is implicated should be psychologically real to the audience. (We will see that this requirement plays a crucial role in their arguments against Grice.) Grice, I will argue, was not pursuing this project. Rather, I will suggest that he was trying to make sense of quite a different notion of what is said: one on which both speaker and audience may be wrong about what is said. On this sort of notion, psychological reality is not a requirement. So objections to Grice based on a requirement of psychological reality will fail. Once Grice's project and that of relevance theorists are seen as distinct, it will be clear that they can happily coexist.2They are simply discussing different subject matters. One may start to wonder, however, about who is really discussing what is said, a topic that both camps claim. I will not attempt a conclusive answer to this question. But I will suggest that Grice's view, despite certain shortcomings, has advantages which seem all too often to have gone unnoticed.Comment: It would make sense to read Grice before engaging with modern reception of his work