PHIL 29909 Philosophy of AI: Language and Large Language Models
An investigation into what philosophy and linguistics can teach us about LLMs, and vice versa. (B)
An investigation into what philosophy and linguistics can teach us about LLMs, and vice versa. (B)
This is a two-part course, taught in the Winter and the Spring. Students may take one part without the other. The first part will be organized around the first half of Stanley Cavell’s major philosophical work, The Claim of Reason. We will focus on the following topics: the role of criteria in epistemology, skepticism about knowledge of the external world, the nature of agreement in judgment, and the relation between meaning and use. In these connections, we will also read work by G. E. Moore, A. J. Ayer, H. H. Price, J. L. Austin, Thompson Clarke, Ludwig Wittgenstein, Barry Stroud, and others.
Issues attending the concepts of negation, limit, and intentionality (construed as thought’s capacity to be answerable to reality) are typically approached in isolation one from another. The course will pursue the contrary hypothesis: namely, that the puzzles arising in connection with these three concepts form a nexus, so that none of them can be comprehended apart from the relations that it entertains with the two others. In order to motivate and substantiate this hypothesis, we will exhume and revive a philosophical tradition that runs from Plato to Wittgenstein through Kant and Sartre and whose defining feature lies in the upholding of this approach. We will examine how the three notions come into play in what Wittgenstein calls “the mystery of negation”: “This is not how things are, and yet we can say how things are not.” Bringing out their nexus requires accounting for the unity and univocity of the concept of negation across two ways of using negation that seem to pull in opposite directions: in the one case, “not-p” makes use of “not” in order to reject p as false (as in “The shirt is not red”), which requires that p lies within the limits of the realm of the intelligible; in the other case, “not-p” makes use of “not” to reject p as nonsensical (as in “The sweet is not a colour”), as if excluding p from the realm of the intelligible. (B)
Readings will include texts by Plato, Maïmonides, Duns Scotus, Spinoza, Kant, Hegel, Brentano, Husserl, Heidegger, Sartre, Frege, and Wittgenstein.
An introduction to philosophical thought about the nature of language. The questions we will address include: What is meaning? What is truth? How does language relate to thought? How do languages relate to each other? What is metaphor? What is fiction? The focus will be on classic work in the analytic tradition (Frege, Russell, Wittgenstein, Carnap, Tarski, Quine, Austin, Grice, Davidson, Donnellan, Putnam, Searle, Kaplan, Kripke) but we will also read, and relate to this modern work, some current work in the philosophical literature and some seminal discussions of language in the writings of Plato and Aristotle. (B)
The observation that ordinary uses of predicates such as “tasty” and “beautiful” trigger an acquaintance inference—they suggest that the speaker has first-hand knowledge of the item under consideration—has received immense attention by philosophers as well as by linguists in recent years. The goal of this seminar is to arrive at a comprehensive and systematic understanding of this phenomenon. We will explore the significance of the acquaintance inference in semantics and philosophy of language (in particular for our understanding of the interaction between literal meaning and discourse pragmatics) but also for aesthetics and meta-ethics. From the linguistics side, we will explore intricate questions surrounding the projection properties of acquaintance inferences as well as issues surrounding “subjective” attitude verbs. The guiding hypothesis of this interdisciplinary seminar is that natural language predicate expressions lexically specify what it takes for their use to be properly ‘grounded’ in a speaker’s state of mind—what state of mind a speaker must be in for a predication to be in accordance with the norms governing assertion—and that these grounding constraints may compositionally interact with other other natural language expressions in interesting ways. (II)
Disjunctivist accounts of human capacities always turn on some form of rejection of (what we will call in this course) a layer-cake assumption. One particularly widespread version of the latter sort of assumption, when asserted as a thesis about the nature of our cognitive faculties and their relation to one another, goes like this: The natures of our sentient and rational capacities respectively are such that we could possess one of these capacities, as a form of cognition, without possessing the other. The underlying assumption is that at least one of these capacities is a self-standing cognitive capacity – one which could operate just as it presently does in us in isolation of the other. This course will begin by examining the counterpart assumption in the philosophy of language, when it is asserted as a thesis about the relation between the aspects of language we respectively apprehend through our power of sensory perception (for example, in recognizing signs) and through our power of intellectual comprehension (for example, in grasping a meaning). One tendency, for example, which we find in much contemporary philosophy of language is to conceive of the linguistic expression as a composite notion to be analyzed in terms of a kind of mere physical mark or acoustic noise to which something further — a meaning or use — is assigned or added in order to yield a fully linguistic expression. Some of the more penetrating philosophers of the past century have noticed that such a conception of language (once it is strictly thought through) appears to encounter insurmountable difficulties. This course will begin by looking at the work of some thinkers in the history of philosophy and linguistics who have challenged such a conception. We will then move on to considering further varieties of layer-cake assumptions and disjunctivist responses thereto that arise in the philosophy of language pertaining to the following further ten interrelated topics: (1) the relation between phonetics and phonology, (2) the relation between phonemes and morphemes, (3) between words and sentences, (4) between infant and adult forms of linguistic capacity, (5) between first and second language acquisition, (6) between orality and literacy in the cultural phases of the historical development of a single natural language, (7) between the pre- and post-punctuation phases in the historical development of the written form of a modern natural language, (8) between the written and spoken sign forms within a single modern natural language, (9) between a logically regimented artificial sign system and a living natural language, and (10) between diverse linguistic forms of speech and/or writing within a single cultural form of life marked by diglossia or heteroglossia. (II)
This course examines the conception of language of the early Wittgenstein though the lens of six common distinctions in the philosophy of language: (1) meaningful sentences vs. meaningful words; (2) semantic content vs. syntactical form; (3) meaningful signs vs. signs; (4) act vs. content; (5) forceful vs. forceless content; and (6) language vs. thought. We will see that the Tractatus challenges familiar ways of construing these distinctions. Specifically, it rejects the view that the second term of each distinction is the conceptually more basic case, while the first term is a composite phenomenon obtained by adding some extra ingredient to the second term. Rather, the second term of each pair, insofar as it is a genuine phenomenon, presupposes in various different ways the other term (sometimes because it is only an abstraction, sometimes because it is a derivative phenomenon, and sometimes because its specification involves derivative notions), or has instead exactly the same status (as in the case, arguably, of language and inner thought). This means that the Tractatus opposes the idea that the full-blown phenomenon of language (that is, language used by some speaker to say something that makes sense) can be reconstructed from a number of more fundamental ingredients. Rather, the full-blown phenomenon of language is the starting point in terms of which each of the aforementioned distinctions, if at all defensible, can be properly vindicated. (B) (IV)
There are no prerequisites for this course, but some previous exposure to the philosophy of language or the history of analytic philosophy is recommended.
The tendency in contemporary philosophy is to conceive of a linguistic sign as a composite notion to be analyzed in terms of kind of mere physical mark or acoustic noise to which something further — a meaning or use — is assigned or added in order yield a meaningful linguistic symbol. This course will explore figures in the history of philosophy and linguistics who opposed such a conception – figures, that is, who thought that the capacity to recognize linguistic signs presupposes some prior comprehension of their real possibilities of use. Readings will be from Frege, Hilbert, early and later Wittgenstein, Franz Boas, Roman Jacobson, Morris Halle, David Kaplan, Sylvan Bromberger, and others. (B) (I)
One previous course in philosophy.
An introduction to philosophical thought about the nature of language. The questions we will address include: What is meaning? What is truth? How does language relate to thought? How do languages relate to each other? What is metaphor? What is fiction? The focus will be on classic work in the analytic tradition (Frege, Russell, Wittgenstein, Carnap, Tarski, Quine, Austin, Grice, Davidson, Donnellan, Putnam, Searle, Kaplan, Kripke) but we will also read, and relate to this modern work, some current work in the philosophical literature and some seminal discussions of language in the writings of Plato and Aristotle.
For each second of John’s life, consider the claim that he is young at that second. Many of these claims will be clearly true: he is young at all of the seconds that make up the first year of his life. Many of these claims will be clearly false: he is not young at all of the seconds that make up his 89th year. If all of these statements are either true or false, it follows that there was a last second at which it is true to say that he is young, and a first second at which it is true to say that he is not young. But that seems wild! One second can’t make the difference between a young person and an old person.
This is one of the central problems raised by the phenomenon of vagueness. This course will examine a variety of philosophical issues raised by the phenomenon of vagueness in the philosophy of language, philosophical logic, epistemology, and metaphysics. Among other things, we will discuss: the philosophical significance of vagueness, the relationship between vagueness and ignorance, decision-making under indeterminacy, and the question of whether vagueness is an essentially linguistic phenomenon. (B)
Elementary Logic (PHIL 20100/30000) or its equivalent.