Scuola di Studi Superiori "Ferdinando Rossi" dell'Università degli Studi di Torino - SSST

The course is an introduction to Artificial Intelligence to a public of students that are not specialists. The main goal is to allow students to recognize the main challenges and tasks related to Artificial Intelligence. Another goal is to offer the opportunity to conduct a practical experience with a software package that will solve specific tasks. The course is divided into four modules. The first one is an introduction to Artificial Intelligence and discusses the classical tasks such as robotics and reasoning mechanisms of an artificial agent. The second module is an introduction to Machine Learning in which we discuss the main tasks of learning from data and the environment by an artificial agent. The third module introduces the methodology of Natural Language Processing, essential to allow an artificial agent to interact by means of the language. The fourth module treats the issues of ethics and law related to Artificial Intelligence and it proposes the discussion on examples of real case stories.

The student is supposed to learn and recognize the main tasks of Artificial Intelligence, their implementation in real use cases, and understand their potentialities and limits.

Module 1 - Introduction to Artificial Intelligence

Teaching staff: Gian Luca Pozzato (10 hours)

• Introduction to Artificial Intelligence
• problems and research fields

- Robotics

• Automated reasoning

- problem solving in the state space search: production rules systems and examples.

 

Module 2 - Machine Learning

Teaching staff: Rosa Meo ( 9 hours)

This module discusses the tasks of Machine Learning, driven from data stored in digital systems.

• An introduction to the tasks of machine learning
• Learning with supervision and in an unsupervised manner.
• Reinforcement learning and by imitation
• Learning that preserves privacy
• We propose an experience in laboratory aided by a software package for learning machine learning models from data (RapidMiner).
• An introduction to artificial neural networks and their variants in deep neural networks with their applications to vision and language
• We treat the problems related to back box models, which are not explainable. We discuss the opportunities and the ingredients  of learning in an explainable way with a trustable model 

 

Module 3 - Natural Language Processing

Teaching staff: Daniele Radicioni (9 hours)

Basic elements on modern technologies for Natural Language Processing will be introduced in this part of the course. The basic levels of analysis will be surveyed: morphology, syntax, semantics and pragmatics. 

Lexical semantic issues will then be introduced, along with some popular computing resources (WordNet and FrameNet, which we call 'cognitively plausible' resources). 

Finally, we will see how text documents can be turned into vector descriptions. We will consider distributed representations that are known as 'word embeddings', and we will explore some use cases where this technology has been successfully employed.

All arguments will be complemented by simple programming assignments to get familiar with the mentioned technologies and theories.

 

Module 4 - Ethics, Law and AI

Teaching staff: Guido Boella

The increasing role of technology in humanity raises constant major challenges to law in a variety of moral, theoretical and doctrinal dimensions. The purpose of this module is to analyze current developments in the fields of Artificial Intelligence ('AI') and robotics through the prism of today’s debate on the ethics of AI and its legal regulation. Along with discussing the interface of law and technology through a variety of critical theoretical perspectives, the module will focus on specific topics and case studies as “laboratories” for assessing contemporary approaches to law & technology. The main objective is to make students aware of the connections between technology, ethics and the legal environment, in order to keep students up-to-date with the current discussions worldwide.

 

Teaching:

The purpose of each class is to address and critically discuss the various topics and case studies under scrutiny. To that end, we envision teaching methods that allow students to actively engage in the proposed discussions, bringing their own national regulatory experiences (or lack thereof) to the table. 

We expect students to prepare for each class, by either reading assigned short texts and/or doing research on proposed topics. Lectures will be based on concrete cases discussions. We will explore both normative theories, such as law and ethics, and the practice behind them. Active student involvement is key to learning in this course. 

The module introduces the state of the art of Artificial Intelligence and ethics.

- Normative impacts that derive from the growing use of AI technology and robotics;

- Law making, ethics and the speed of innovation;

- Practical cases for introductory analysis: An overview of our case studies.

- It discusses AI agenthood.

- The case study of Sophia, i.e. the first AI application to receive citizenship of any country (Saudi Arabia, October 2017);

- Agents and persons: The doubts of the EU Commission and of the EU Parliament;

- What should we expect in the foreseeable future?

It discusses the relationships between AI and human creativity.

- AI and the economics of knowledge

- The frontiers of today’s intellectual property

- AI as ‘authors’

 It concludes with presentation of the problems related to a new generation of crimes.

- The case study of Vital and Alicia T. (two robo-traders); 

- 25 years of computer crimes: what we have learned so far, and what's next;

- The laws of war.

Lectures and workshops

A short written assignment (maximum 2500 words excluding bibliography) on one of the topics covered in the lectures, or on a topic agreed with one of the lecturers, is required as a final assessment of achievement of the learning objectives.