The Philosophy of Artificial Intelligence
Syllabus and slides
While references to Artificial Intelligence (AI) are pervasive in popular culture and in the technology industry, the phrase is rarely defined. What is AI? Is a thermostat an AI system? What about digital assistants like Siri and Alexa? Or DeepBlue, Watson, and AlphaGo – the first computer programs to beat humans at the games of chess, Jeopardy!, and go, respectively? Does AI even exist today? If not, will it ever exist, or is it an impossible project? Raising these deceptively simple questions takes us into deep philosophical territory. Despite some AI researchers' professed disinterest for philosophy, the project of building AI is in fact intimately related to philosophical questions, so much so that it is often difficult to disentangle strictly "technical" issues from philosophical ones.
This course will explore various questions at the intersection between philosophy and AI. We will start by tracing the deep philosophical roots of the AI project. This historical overview will lead us to ask how we should define AI in the first place, which in turn raises interesting questions about the nature of intelligence in biological organisms and artificial systems. After distinguishing several notions of AI, we will critically examine influential arguments against the possibility of "Strong AI" – the project of building artificial persons that have the same mental capacities as humans.
We will also discuss concrete approaches to building AI: logic-based approaches, that rely on explicit rules and symbolic representations, and connectionism, that rely on artificial neural networks inspired by the brain. Taking a closer look at several active areas of AI research, we will explore their connections to philosophy: computer vision (philosophy of perception) and natural language processing (philosophy of mind and language).
Finally, we will explore ethical and aesthetic implications of AI research. The hypothetical prospect of creating superhuman artificial intelligence has prompted concerns about the "existential risk" that AI research might pose for the future of humanity. While some philosophers take these concerns very seriously, others dismiss them as little more than science-fiction. The latter emphasize that we should focus on more pressing ethical concerns regarding current and near-term uses of AI. A first set of applied issues regards the deployment of physical autonomous systems that will kill humans, either accidentally (e.g., autonomous vehicles) or by design (e.g., autonomous weapons). A second set of issues regards fairness and bias in AI systems used to curate data (e.g., recommendation engines, or social media feed curation), or assist decisions that have a significant impact of people's lives (e.g., predicting repeat offenses, creditworthiness, or grades). The production of artworks made with algorithms also raises fascinating questions related to the philosophy of art.
- Week 1 – Philosophy and AI: General Overview
- Week 2 – The Philosophical Foundations of AI
- Week 3 – Varieties of (Artificial) Intelligence
- Week 4 – The Case Against Strong AI
- Week 5 – Symbolic and Connectionist AI
- Week 6 – Computer Vision and Natural Language Processing
- Week 7 – Opening the Black Box: Opacity and Explanation in AI
- Week 8 – Feedback on first paper & class discussion
- Week 9 – The Ethics of AGI
- Week 10 – AI Ethics in the Real World (i): Autonomous systems
- Week 11 – AI Ethics in the Real World (ii): Fairness and Bias
- Week 12 – AI Aesthetics