In a recent letter, Dillion et. al (2023) make various suggestions regarding the idea of artificially intelligent systems, such as large language models, replacing human subjects in empirical moral psychology. We argue that human subjects are in various ways indispensable.

As machine learning informs increasingly consequential decisions, different metrics have been proposed for measuring algorithmic bias or unfairness. Two popular “fairness measures” are calibration and equality of false positive rate. Each measure seems intuitively important, but notably, it is usually impossible to satisfy both measures. For this reason, a large literature in machine learning speaks of a “fairness tradeoff” between these two measures. This framing assumes that both measures are, in fact, capturing something important. To date, philosophers have seldom examined (...) this crucial assumption, and examined to what extent each measure actually tracks a normatively important property. This makes this inevitable statistical conflict – between calibration and false positive rate equality – an important topic for ethics. In this paper, I give an ethical framework for thinking about these measures and argue that, contrary to initial appearances, false positive rate equality is in fact morally irrelevant and does not measure fairness. (shrink)

On a traditional view of perceptual justification, perceptual experiences always provide prima facie justification for beliefs based on them. Against this view, Matthew McGrath and Susanna Siegel argue that if an experience is formed in an epistemically pernicious way then it is epistemically downgraded. They argue that "wishful seeing"—when a subject sees something because he wants to see it—is psychologically and normatively analogous to wishful thinking. They conclude that perception can lose its traditional justificatory power, and that our epistemic norms (...) should govern how experiences are formed. To make this case, the downgrader must first isolate a feature of wishful thinking that makes it epistemically defective, then show that this feature is present in wishful seeing. I present a dilemma for the downgrader. There are two features of wishful thinking that could plausibly explain why it is irrational: the fact that a desire causes you to form a belief not supported by adequate evidence, or the mere influence that desire holds over belief formation. Each option presents formidable difficulties. Although the first “bad evidence” explanation, which McGrath employs, explains the irrationality of wishful thinking, it does not transfer to wishful seeing, since experiences are not formed in response to evidence. The second “influence of desire” explanation, which Siegel employs, fails to isolate an epistemically defective feature of wishful thinking, and also does not transfer to wishful seeing. I conclude that the downgrader’s argument from wishful seeing fails. (shrink)

This paper considers the kind of introspection that large language models (LLMs) might be able to have. It argues that LLMs, while currently limited in their introspective capabilities, are not inherently unable to have such capabilities: they already model the world, including mental concepts, and already have some introspection-like capabilities. With deliberate training, LLMs may develop introspective capabilities. The paper proposes a method for such training for introspection, situates possible LLM introspection in the 'possible forms of introspection' framework proposed by (...) Kammerer and Frankish, and considers the ethical ramifications of introspection and self-report in AI systems. (shrink)

Historically, the dispute between empiricists and nativists in philosophy and cognitive science has concerned human and animal minds (Margolis and Laurence in Philos Stud: An Int J Philos Anal Tradit 165(2): 693-718, 2013, Ritchie in Synthese 199(Suppl 1): 159–176, 2021, Colombo in Synthese 195: 4817–4838, 2018). But recent progress has highlighted how empiricist and nativist concerns arise in the construction of artificial systems (Buckner in From deep learning to rational machines: What the history of philosophy can teach us about the (...) future of artificial intelligence. Oxford University Press.). This paper uses nativism and empiricism to address questions about the nature of artificial intelligence and its trajectory. It begins by defining the nativism/empiricism debate in terms of the generality of a system. Nativist systems have initial states with domain-specific features; empiricist systems have initial states with only domain-general features. With the debate framed in this way, it then explores a variety of arguments for nativism and empiricism in AI. These arguments revolve around two different questions which must be distinguished: whether nativism the only possible approach to developing human-level AI (HLAI); and whether nativism is the most practical approach to developing HLAI. On the first question, it argues that nativism is quite clearly not the only possible approach to developing HLAI, as is sometimes suggested. It argues that existing arguments for the necessity of nativism are unconvincing, because they analogize from poverty of the stimulus arguments about humans, while AIs often have access to much more data than humans. Then it argues that the case of evolution gives us a compelling argument against nativism. On the second, practical question, the paper argues that there is a tradeoff between the advantages of encoding innate machinery directly, and the advantages of evolving or learning it. However, as the past decade has shown, empiricism is a much more viable path to greater capability levels, given the ‘bitter lesson’ (Sutton in _Reinforcement Learning: An Introduction_. MIT press.) that encoding the ‘correct’ knowledge in AI systems is perennially outperformed by more empiricist methods that leverage large-scale data and computation. (shrink)