75943 - Artificial Intelligence, Problem solving and the Semantic Web (1)

Course contents

Artificial Intelligence: Philosophy, Ethics and Society

 

SYLLABUS available here

This course introduces the ethical, social, environmental and philosophical problems posed by artificial intelligence. In particular, the course will focus on the advent of generative artificial intelligence, which with LLM models such as ChatGPT has changed the landscape of artificial intelligence.

Two questions in particular will be the focus of the course:

Q1) Can artificial intelligence result in artificial beings that are as intelligent as or more than human beings?

Q2) What challenges does artificial intelligence pose for the responsible use of digital technologies?

Course structure

The course is divided into three parts:

1) fundamentals of artificial intelligence

In this first part, the basics of artificial intelligence will be introduced, which are necessary to address the topics of the second and third parts. The exposition will be as non-technical as possible and will not presuppose any knowledge of computer science.

1.1 algorithm

1.2 symbolic artificial intelligence

1.3 sub-symbolic artificial intelligence:

1.3.1 machine learning

1.3.2 neural network

1.3.3 Large Language Models (LLM) and the case of ChatGPT (the transformer technology

2) The philosophical foundations of artificial intelligence

This part will present some issues raised in the philosophical debate concerning the relationship between human and artificial cognition.

2.1 the Turing Test

2.2 Searle's Chinese room argument

2.3 Consciousness and LLM: Chalmers

3) Ethical, social and environmental problems posed by artificial intelligence

In this final section, some of the problems raised by artificial intelligence will be analysed.

3.1 The problem of value alignment

3.2 AI and existential risks

3.3 Deskilling and the problem of potential job losses replaced by artificial intelligence

3.4 Algorithmic biases: transparency, explainability and epistemic injustice

3.5 The environmental impact of computational architectures

Readings/Bibliography

The course will be based on parts of the following books:

rawford, Kate. Né intelligente né artificiale Il lato oscuro dell'IA. Bologna: Il Mulino, 2021.

Boden, Margaret. L'intelligenza artificiale. A cura di Diego Marconi. Bologna: Il Mulino, Universale paperbacks Il Mulino, 2019.

Cristianini, Nello. Machina sapiens. Bologna: Il Mulino, 2024.

Giunti, Marco, Pinna, Simone e Garavaglia, Fabrizia Giulia. Menti e macchine. Teorie filosofiche e scientifiche, Milano: Le Monnier Università, 2022.

Floridi, Luciano. Etica dell’intelligenza artificiale. Sviluppi, opportunità, sfide, Milano: Raffaello Cortina, 2022.

Galletti, Matteo e Zipoli Caiani, Silvano (a cura di). Filosofia dell'Intelligenza Artificiale. Sfide etiche e teoriche, Bologna: Il Mulino, 2024.

Giunti, Marco, Pinna, Simone e Garavaglia, Fabrizia Giulia, Menti e macchine. Teorie filosofiche e scientifiche, Milano: Le Monnier Università, 2022.

O'Neil, Cathy. Armi di distruzione matematica. Milano: Bompiani, 2017.

****

The academic literature on artificial intelligence is vast and has seen rapid growth in recent years. As is typical in scientifically mature disciplines, the best literature is published in English in international specialized journals. Besides these texts, the following English texts, whose reading is optional, will serve as a background for the lessons:

Can machines think?

• Turing, A. M. (1950) Computing machinery and intelligence. Mind, 49, 433-460. Link [https://www.cs.mcgill.ca/~dprecup/courses/AI/Materials/turing1950.pdf]

• Searle, John. R. (1980) Minds, brains, and programs. Behavioral and Brain Sciences, 3(3): 417-457. Link

• Chalmers, David J. (2023) Could a Large Language Model Be Conscious? Boston Review, August 9, 2023. Link [https://www.bostonreview.net/articles/could-a-large-language-model-be-conscious/]

The moral status of advanced artificial intelligences:

• Schwitzgebel, Eric, and Mara Garza (2020) Designing AI with Rights, Consciousness, Self-Respect, and Freedom, in S. Matthew Liao (ed.), Ethics of Artificial Intelligence, New York, Oxford University Press.

• Müller, V.C. (2021) Is it time for robot rights? Moral status in artificial entities. Ethics and Information Technology, 23:579–587.

• Véliz, C. (2021) Moral zombies: why algorithms are not moral agents. AI & Society 36: 487–497.

• Liao, S. Matthew (2020) The Moral Status and Rights of Artificial Intelligence, in S. Matthew Liao (ed.), Ethics of Artificial Intelligence, New York, Oxford University Press.

Artificial General Intelligence/Superintelligence, existential risk, and the value alignment problem.

• Russell, S. (2020) Artificial Intelligence: A Binary Approach. Oxford University Press, in S. Matthew Liao (ed.), Ethics of Artificial Intelligence, New York, Oxford University Press. p. 327-341.

• Bengio, Yoshua. (2023) AI and Catastrophic Risk, Journal of Democracy, September 2023, Link

• Russell, Stuart J. (2019) Human Compatible: Artificial Intelligence and the Problem of Control. New York: Viking.

• Bostrom, Nick. (2014) Superintelligence: paths, dangers, strategies. Oxford University Press.

• Bengio, Yoshua. FAQ on Catastrophic AI Risks. yoshuabengio.org [https://yoshuabengio.org/2023/06/24/faq-on-catastrophic-ai-risks/]

• Maclure, J. (2020) The new AI spring: a deflationary view. AI and Society, 35: 747-750.

Deep Learning & Large Language Models

• Shanahan, Murray. Talking About Large Language Models. December 7, 2022. Link [https://arxiv.org/abs/2212.03551]

• Floridi, L. (2023) AI as Agency Without Intelligence: on ChatGPT, Large Language Models, and Other Generative Models. Philosophy and Technology. 36,15.

• Bubeck et al. Sparks of Artificial General Intelligence; Early experiments with GPT-4. March 22, 2023. Link [https://arxiv.org/abs/2303.12712]

• LeCun, Y., Bengio, Y. & Hinton, G. (2015) Deep learning. Nature. 521, 436–444.

• Buckner, C. (2019) Deep learning: A philosophical introduction. Philosophy Compass, 14(10), 1-19.

• Bender, E., Gebru, T. et al. (2021) On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? FAccT '21: Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, 610-623.

• Choi, Yejin. (2022) The Curious Case of Commonsense Intelligence. Daedalus 151 (2): 139–155.

Designing virtuous artificial agents

• Wallach, W., Vallor, S. (2020) Moral Machines: From Value Alignment to Embodied Virtue, in S. Matthew Liao (ed.), Ethics of Artificial Intelligence, New York, Oxford University Press, ch. 13, 383-412.

• Rini, Regina. Creating Robots Capable of Moral Reasoning Is like Parenting: Aeon Essays. Aeon, 2017. Link [https://aeon.co/essays/creating-robots-capable-of-moral-reasoning-is-like-parenting]

Automation, distributive justice, and the meaning of work

• Danaher, John (2017) Will Life Be Worth Living in a World Without Work? Technological Unemployment and the Meaning of Life. Science and Engineering Ethics 23 (1):41-64.

• Nieswandt, Katharina (2021) Automation, Basic Income and Merit. In Keith Breen & Jean-Philippe Deranty (eds.), Whither Work? The Politics and Ethics of Contemporary Work. Milton and New York: Routledge. pp. 102–119.

• James, A. (2020) Planning for Mass Unemployment, in S. Matthew Liao (ed.), Ethics of Artificial Intelligence, New York, Oxford University Press, ch. 6, 183-211.

The problem of explainability in artificial intelligence

• Buckner, Cameron (2023) Black Boxes or Unflattering Mirrors? Comparative Bias in the Science of Machine Behaviour. British Journal for the Philosophy of Science 74 (3):681-712.

• Maclure, J. (2021) AI, Explainability and Public Reason: The Argument from the Limitations of the Human Mind. Minds and Machine, 31, 421-438.

• Zerilli, J., Knott, A., Maclaurin, J., & Gavaghan, C. (2019) Transparency in algorithmic and human decision-making: Is there a double standard? Philosophy & Technology, 32(4), 661–683.

• Vredenburgh, Kate (2021) The Right to Explanation. Journal of Political Philosophy 30 (2):209-229.

• Vaassen, B. (2022) AI, Opacity, and Personal Autonomy. Philosophy and Technology 35, 88.

Predictive algorithms, recommendation systems, autonomy, and privacy

• Prunkl, C. (2022) Human Autonomy in the Age of Artificial Intelligence. Nature Machine Intelligence 4 (2):99-101.

• Yeung, Karen. (2017) Hypernudge: Big Data as a mode of regulation by design. Information, Communication & Society, 20:1, 118-136.

• Laitinen, Arto & Sahlgren, Otto (2021) AI Systems and Respect for Human Autonomy. Frontiers in Artificial Intelligence 4.

• Jesse, Mathias & Jannach, Dietmar. (2021) Digital nudging with recommender systems: Survey and future directions. Computers in Human Behavior Reports 3.

Ethics and artificial intelligence

• Susser, D., Roessler, B., Nissenbaum, H. (2019) Online Manipulation: Hidden Influences in a Digital World. 4 Georgetown Law Technology Review 1.

• Mhlambi, S., Tiribelli, S. (2023) Decolonizing “AI Ethics: Relational Autonomy as a Means to Counter AI Harms.” Topoi 42, 867–880.

• van Maanen, G. (2022) AI Ethics, Ethics Washing, and the Need to Politicize Data Ethics. DISO 1, 9.

• Birhane, A. (2021) Algorithmic injustice: a relational ethics approach. Patterns 2(2):100205.

• Russo, F., Schliesser, E. & Wagemans, J. (2023) Connecting ethics and epistemology of AI. AI and Society: 1-19.

• Cole M., Cant C., Ustek Spilda F., Graham M. (2022) Politics by Automatic Means? A Critique of Artificial Intelligence Ethics at Work. Frontiers in Artificial Intelligence 5.

• Seger, E. (2022) In Defence of Principlism in AI Ethics and Governance. Philosophy and Technology 35.

Teaching methods

Active learning methodology

This teaching seeks to implement an active teaching method aimed at facilitating learning in a context of group activities.

To do this, three actions will be implemented

1. the use of peer instruction during lessons;
2. the use of the Wooclap application (http://wooclap.com);
3. the use of the Perusall social reading platform (http://perusall.com).

This course will have an optional blended format in which of in addition to face-to-face lectures there will be activities in asynchronous online on the Perusall platform (Perusall.com).

TEACHING METHODOLOGY

The methodology employed is that of peer instruction invented by Harvard experimental physicist Eric Mazur (https://en.wikipedia.org/wiki/Peer_instruction ). Peer instruction is a teaching method that exploits the potential of social interaction to guide learning. Synchronous face-to-face lectures will take place by means of a series of in-class comprehension questions -ConcepTests- to test understanding of specific points of the course material, students will answer these questions and work in small groups confronting each other when it comes to addressing the points that are most difficult for them. The lecturer will play both the role of knowledge transmitter when it comes to clarifying some misunderstandings in relation to the questions, and the role of facilitator when it comes to facilitating the discussion of the questions in groups. Each lesson will presuppose the reading of materials that will be read and discussed by the students beforehand online in the digital asynchronous social learning environment provided by the Perusall.com platform (specially designed by Mazur's Harvard group to support peer instruction).

During in-class hours, students will be able to answer questions posed by the lecturer using the free Wooclap application (http://wooclap.com/ ). Depending on the answers given, the teacher will or will not ask the students to discuss with each other for a few minutes and then check their understanding of the point again. The answers given in class will be stored anonymously in the system and will not be assessed for the final examination.

The conduct of an in-class lesson using peer instruction can be illustrated with the following steps:

1. Students read the assigned materials on the Perusall social reading software (https://perusall.com/) and leave questions and comments before coming to class.

• Students will have access to the online course materials via both Virtual and Perusall (https://perusall.com/ ).
• Attending students will be asked to read the assigned material on Perusall once a week and make comments and questions online (especially on points that are not clear to you).

2. The lecturer reviews the students' feedback on the texts read in advance on Perusall by interacting with them asynchronously.

• During face-to-face class time, students' misconceptions and difficulties that arose online through students' questions on Perusall are elicited, addressed and resolved.
• A lesson is divided into a series of comprehension questions -ConcepTest- possibly preceded by mini-lessons.
• A ConcepTest is a short conceptual question designed to give students an opportunity to test their learning (student observations on Perusall are the best material for creating ConcepTests).
• Wooclap (https://app.wooclap.com/) will be used to have the ConcepTests answered during the lesson.

3. The structure of a face-to-face lesson is as follows:

• A lesson topic can be presented with a mini-lesson (10-15 minutes) or you go directly to the question.
• A ConcepTest is presented.
• Students first answer the ConcepTest individually.
• If most students provide incorrect answers, students are asked to discuss their answers in small groups with their peers and instructors, and then answer again.
• The cycle is completed with an activity to clarify any incorrect answers through guided class discussion.

Assessment methods

The final grade in the examination will be based on the evaluation of a written essay (also called 'paper') and an oral examination in which the essay will be discussed.

The length of the essay is:

(1) for those who have enrolled in Perusall and done all the assignments (attending students) short essay;
(2) for those who have not enrolled in Perusall or have not done all the assignments long essay (non-attending students).

Short essay length (for those who have enrolled in Perusall and have done all 4 assignments): at least 1000 words and no more than 2000 words (all inclusive: first name, surname, course of study, title, bibliography).

Length of long essay (for those who have not enrolled in Perusall or have not done all 4 assignments): at least 1500 words and no more than 2500 words (all inclusive: first name, surname, freshman, course of study, title, bibliography).

Formatting: double line spacing, font size 12
Electronic format pdf, doc or odt.

Essay topic: The choice of topic must be within the course content.

The reference bibliography consists of the texts covered in the course together with any secondary literature to be found in the syllabus bibliography that will be provided on Virtuale.

Submission:The essay must be submitted online on Compilatio (the link will be given on Almaesami at the same time as the roll call).

ERASMUS STUDENTS: erasmus students can write the essay in English and can make use of an equivalent bibliography in English.

EVALUATION CRITERIA

I will use these criteria to determine the following assessment thresholds:

30 and praise excellent proof, both in knowledge and in the critical and expressive articulation.

30 excellent test, complete knowledge, well articulated and correctly expressed, with some critical ideas.

27-29 good test, comprehensive and satisfactory knowledge, substantially correct expression.

24-26 discrete test, knowledge present in the substantial points, but not exhaustive and not always correctly articulated.

21-23 sufficient proof, knowledge present in a sometimes superficial way, but the general thread is understood. Short and often inappropriate and incomplete expression and articulation.

18-21 superficial knowledge, the common thread is not understood with continuity. The expression and the articulation of the discourse also have significant gaps.

<18 insufficient evidence, absent or very incomplete knowledge, lack of orientation in the discipline, defective and inappropriate expression. Examination not passed.

Students with disabilities and Specific Learning Disorders (SLD)

Students with disabilities or Specific Learning Disorders are entitled to special adjustments according to their condition, subject to assessment by the University Service for Students with Disabilities and SLD. Please do not contact teachers or Department staff, but make an appointment with the Service. The Service will then determine what adjustments are specifically appropriate, and get in touch with the teacher. For more information, please visit the page: https://site.unibo.it/studenti-con-disabilita-e-dsa/en/for-students

Teaching tools

Virtuale, Wooclap, Perusall

Office hours

See the website of Sebastiano Moruzzi