Petrus Hispanus Lectures 2026
School of Arts and Humanities, University of Lisbon
April 28 2026 16h-18h – Sala A201 Anfiteatro III
April 29 2026 16h-18h – Sala B112.D (near the Library of the Faculty of Arts and Humanities)
Against Intelligence
Tim Crane, Central European University
ABSTRACT
These lectures aim to answer a simple question arising from Alan Turing’s famous 1950 paper, ‘Computing Machinery and Intelligence’, and to draw some conclusions from this answer. Turing’s paper is often taken as one of the foundational texts of the discipline of Artificial Intelligence (AI), whose birth is usually dated from 1956. But what does Turing actually say about intelligence? Almost nothing: the word ‘intelligence’ occurs only twice in the paper, and ‘intelligent’ only once. So what was Turing really talking about, if not intelligence? And what has this got to do with today’s AI?
LECTURE 1: WHAT IS INTELLIGENCE?
Many recent theoretical discussions of AI have centred around the possibility of Artificial General Intelligence (AGI) — an artificial version of the kind of open-ended, general-purpose intelligence that humans are supposed to have. But what exactly is this? AI researchers have offered a number of definitions of AGI: on some of these definitions, AGI has already been achieved, and on others it is still far beyond our reach. What is going on? I argue that the concept of intelligence does not demarcate a cognitive capacity, but rather serves primarily as an evaluative concept, used to classify exercises of cognitive capacities. I show how this way of thinking resolves a number of puzzles about AGI and so-called ’Superintelligence’.
LECTURE 2: WHAT CAN COMPUTERS DO?
Lecture 1 gave the solution to the Turing puzzle: Turing was not really interested in intelligence, but in whether computers can think, and how we could tell. This solution is not just of historical or scholarly interest, but has significant consequences for today’s AI. It makes sense for AI researchers to evaluate the performance of machines as intelligent or not, but it does not make sense for them to investigate the nature of intelligence as such — in the relevant sense, there is no such thing. The question rather should be, what kinds of cognitive tasks can computers perform? This question is both more tractable that the question about intelligence, and more open-ended. I will argue that the discussion of AGI in Lecture 1 gives us the clue how to proceed: every computation involves a specific task, and where there is no task there can be no computation. I apply these ideas to recent speculations about AI consciousness.
The lectures will close with some proposals about how the philosophy of AI can help us with the practical problems that contemporary AI has created.
This event is funded by Portuguese national funds through FCT – Fundação para a Ciência e a Tecnologia, I.P., within the project UID/00310/2025, Centro de Filosofia da Universidade de Lisboa (https://doi.org/10.54499/UIDB/00310/2025)