Philosophy and Artificial Intelligence 2023
Jobst Landgrebe and Barry Smith
MAP, USI, Lugano, Spring 2023
Background
Much of the material for this class is derived from our book Why Machines Will Never Rule the World: Artificial Intelligence without Fear (Routledge 2022). The March 1-2 sessions of the class will be accompanied by afternoon sessions devoted to a conference on the book.
Table of contents, first chapter and references
Faculty
Jobst Landgrebe is the founder and CEO of Cognotekt, GmBH, an AI company based in Cologne specialised in the design and implementation of holistic AI solutions. He has 17 years experience in the AI field, 8 years as a management consultant and software architect. He has also worked as a physician and mathematician.
Barry Smith is one of the world's most widely cited philosophers. He has contributed primarily to the field of applied ontology, which means applying philosophical ideas derived from analytical metaphysics to the concrete practical problems which arise where attempts are made to compare or combine heterogeneous bodies of data.
Course Description
Artificial Intelligence (AI) is the subfield of Computer Science devoted to developing programs that enable computers to display behavior that can (broadly) be characterized as intelligent. On the strong version, the ultimate goal of AI is to create an artificial system that is as intelligent as a human being. Recent striking successes such as AlphaFold have convinced many not only that this objective is obtainable but also that in a not too distant future machines will become even more intelligent than human beings.
The actual and possible developments in AI open up a series of striking questions such as:
- Can a computer have a conscious mind?
- Can a computer have desires, a will, and emotions?
- Can a computer have responsibility for its behavior
- Would machine intelligence, if there is such a thing, be something comparable to human intelligence or something quite different?
In addition, new developments in the AI field make it possible for us to consider a series of philosophical questions in a new light, including:
- Could a machine have something like a personal identity? Would I really survive if the contents of my brain were uploaded to the cloud?
- What is it for a human to behave in an ethical manner? (Could there be something like machine ethics? Could machines used in fighting wars be programmed to behave ethically?)
- What is a meaningful life? If routine, meaningless work in the future is performed entirely by machines, will this make possible new sorts of meaningful lives on the part of humans?
After introducing the relevant ideas and tools from both AI and philosophy, all the aforementioned questions will be thoroughly addressed in class discussions. The class will close with presentations of papers on relevant topics given by students.
Draft Schedule
Tuesday, Feb 28 (15:30-18:00pm) Why machines will never rule the world
- Barry Smith: Overview of the arguments and introduction to the human mind.
- Room:
- Announcement: Why Machines Will Never Rule the World
Introduction to the class
Readings:
- John Searle: Minds, Brains, and Programs
- Jobst Landgrebe and Barry Smith: There is no Artificial General Intelligence
- Max More and Natasha Vita-More (Eds.), The Transhumanist Reader: Classical and Contemporary Essays on the Science, Technology, and Philosophy of the Human Future, Wiley-Blackwell, 2013.
Wednesday March 1 (9:30-12:00 noon): Artificial General Intelligence and Introduction to Stochastic AI
- Room:
Artificial General Intelligence
Bibliography of Why Machines Will Never Rule the World
The classical psychological definitions of intelligence are:
- A. the ability to adapt to new situations (applies both to humans and to animals)
- B. a very general mental capability (possessed only by humans) that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly, and learn from experience
What are the essential marks of human intelligence?
Readings:
- Linda S. Gottfredson. Mainstream Science on Intelligence. In: Intelligence 24 (1997), pp. 13–23.
The Legg-Hutter Definition of Intelligence
What is it that researchers and engineers are trying to do when they talk of achieving ‘Artificial Intelligence’?
To what extent can AI be achieved?
Problems with the Legg-Hutter Definition of Intelligence
Theory of complex systems
Readings:
- Shane Legg and Marcus Hutter: Universal Intelligence: A Definition of Machine Intelligence
- Jobst Landgrebe and Barry Smith: Making AI Meaningful Again
- S. Thurner et al. (2018): Introduction to the theory of complex systems (Oxford):
Wednesday March 1 (14:30-18:00): Part 1 of Conference on Why Machines Will Never Rule the World
14:30 – 14:45 Welcome
14:45 – 15:15 Barry Smith (Buffalo, USI): Summary of the Book
Video of Opening Presentation by Landgrebe and Smith
15:15 – 15:30 Jobst Landgrebe (Buffalo, USI) "The Core Mathematical Argument"
15:30 –16:00 Landgrebe: "Large Language Models (Galactica, ChatGPT)", Smith: "ChatGPT heuristics" (Examples of ChatGPT hallucinations)
16:00 – 16:15 Q&A
16:15 – 16:30 Coffee Break
16:30 – 18:00 Tim Crane (CEU, Vienna) “Is Artificial General Intelligence Possible?”
Thursday, March 2 (9:30-12:00 noon): Language and Natural Intelligence
Language Structure and Language Economy (Landgrebe)
Large Language Models (Landgrebe)
Natural Intelligence (Landgrebe)
Thursday March 2 (14:00-18:00): Part 2 of Conference on Why Machines Will Never Rule the World
14:30 – 16:00 Emma Tieffenbach (USI, Zürich) “Making Sense of Singularity”
16:00 – 16:15 Coffee Break
16:15 – 17:45 Stefan Wolf (USI) “agAInst”
17:45 – 18:15 Concluding replies by Jobst Landgrebe & Barry Smith
Friday, March 3 (13:30 - 16:00pm) Digital Immortality and the Meaning of Life
Tuesday, May 9 (15:30 - 18:00pm) The Machine Will
The Turing test
What is the mind?
Can machines have a consciousness?
- Searle’s wall
- The Lucas-Penrose argument against machine consciousness
- Bringsjord’s defence of machine consciousness
- Other approaches in philosophy of mind, computation and AI
- Computational theory of the mind
Objectifying intelligence and theoretical thinking
- Why machines will not master human language
- Neural Machine Translation
- Human language as a complex system
- Properties of the language system
- AI conversation emulation
- Challenges to machine conversation-->
- Initial utterance production
- Why machines will not master human language
- Modelling dialogue dynamics mathematically
- Mathematical models of human conversations
- Current state-of-the-art in dialogue systems
- Why conversation machines are doomed to fail
- Chapter 11 Why machines will not master social interaction 224
- No AI emulation of social behaviour
- Some examples
- No machine intersubjectivity
- No machine social norms
- AI and legal norms
- No machine emulation of morality
- No explicit ethical agents
- No AGI in the kill chain
What is will?
Can machines have a will?
Why no Robot Cops?
Could a machine have goals?
Questions
Readings:
- Moor: Four kinds of ethical robots
- Crane: The AI Ethics Hoax
Wednesday, May 10 (9:30 - 13:00) Logic and Complex Systems: Computation
- Implementing mathematical models in computers
- Classical computation
- Turing machines
- Classical logic-gate based computation
- Quantum computation
- Quantum mechanics
- Quantum computing
- Why quantum computers are Turing machines
- Classical computation
Thursday, May 11 (9:30 - 12:00 noon) Logic and Complex Systems
- The Limits of Mathematical Models
- Slides
- Models
- All science requires mathematical models
- Types of models 1: descriptive, explanatory, predictive
- Types of models 2: qualitative, quantitative
- All predictive models are quantitative
- Synoptic models
- Adequate models
- Computability
- All AI engineering requires mathematical models
- Explicit and implicit mathematical models
- Systems
- System elements and system interactions
- Systems are fiat entities: they are a product of delimitation
- System boundaries
- Relatively isolated systems
- The Limits and Potential of AI
- Intentions and drivenness
- No emulation of animate drivenness
- Models
- AI and the Mathematics of Complex Systems
- Slides
- Bayesian networks
- Complex systems
- Comprehensive and partial models
- The scope of extended Newtonian mathematics
- Seven Properties of complex systems
- Examples of complex systems
- Human beings as complex systems
- Complex systems of complex systems
- Animate complex systems are organized and stable
- Mathematical models of complex systems
- Multivariate distributions
- Adequate models for complex systems
- Predictive models of complex systems
- Why we ain’t rich
- Example of a social fact
- Why we ain’t rich
- Approaches to complex system modelling
- Naïve approaches
- Consequences for AI applications
- Refined approaches
- Scaling
- Explicit networks
- Evolutionary process models
- Entropy models
- Complex system emulation requires complex systems
- AI and the Ontology of Power, Social Interaction and Ethics
- Preliminary Video
Tuesday May 16 (15:30 - 18:00pm): The Nature of Intelligence: Humans vs. ChatGPT
Human and machine intelligence
Can ChatGPT become intelligent?
Are Large Language Models a threat to humanity?
Capabilities, or: What do IQ tests measure?
Is Psychology Finished?
Wednesday May 17 (9:30 - 12:00 noon) Student Presentations and Concluding Survey
- Student Presentations
Aaron Wirt: Politics and Digital Technologies
Sven Eichholtz
Background Reading
An Introduction to AI for Philosophers
(AI experts are invited to criticize what I have to say here)
An Introduction to Philosophy for Computer Scientists
(Philosophers are invited to criticize what I have to say here)
