Objectives, program and content of the course, focused on artificial intelligence, promoted by Collegio Nuovo and included in the three-year degree program in Communication, Innovation, Multimedia (CIM) at the University of Pavia
Big data, machine learning and artificial intelligence: knowledge, information, power is an elective course included in the 3rd year of the Bachelor of Arts in Communication, Innovation, Multimedia (CIM) program at the University of Pavia. The teaching, promoted by the Collegio Nuovo, is part of the SPS-08 (Sociology of Cultural and Communicative Processes) area and gives rise to 6 CFUs. For the Italian version of this page, see here.
1. Prerequisites
The course does not require prior IT skills or programming knowledge, nor does it have any formal prerequisites. Since it is offered as an elective course and is open to students from different academic backgrounds, the knowledge and skills expected upon entry are described in full below.
Required Knowledge
– Basic descriptive statistics. Students should be able to recognise and interpret: variables and distributions; arithmetic mean, median and mode, and the circumstances in which one is more informative than another; standard deviation as a measure of dispersion; correlation and its distinction from causation; samples and inference. The ability to perform calculations is not required; students are expected to be able to recognise the appropriate — and inappropriate — use of these concepts in a journalistic text, market research report or corporate document.
– B2-level reading comprehension in English according to the CEFR. Some of the recommended further reading, the documentation for the platforms used in the laboratory sessions, and one of the two monographs prescribed for Examination Path B are in English.
Required Skills
– Basic digital skills at user level: managing files and folders, proficient use of a web browser, drafting a text document, preparing a presentation, and independently creating and managing an account on a web platform.
– Ability to read and summarise an argumentative text of moderate complexity and to independently locate a source on the basis of a bibliographic reference.
Required Tools
For the laboratory component, a free account on the platforms indicated at the beginning of the course is sufficient. No paid subscription is required, and none of the exercises relies on features available only through paid plans.
Bridging Prerequisite Knowledge
Students who feel they need to consolidate their knowledge of statistics are advised to read the introductory chapters of one of the textbooks listed under “Reference Texts”. During the first week of the course, an anonymous, non-assessed self-evaluation test (ten questions, ten minutes) will be available on KIRO to help students identify which topics they may need to review. The test has no bearing whatsoever on the final assessment.
2. Learning Objectives and Outcomes
The course aims to provide students with the conceptual tools required to understand how the numerical representation of the world and the principle of computability are reshaping the production of information, communication markets and forms of power, as well as the practical ability to use generative artificial intelligence systems competently and transparently.
2.1 Knowledge and Understanding
By the end of the course, students will be able to:
– reconstruct the main stages in the history of the idea of the computability of problems in philosophy and mathematics, from Euclid to Alan Turing, and place the emergence of computer science and artificial intelligence within this historical framework;
– correctly define and distinguish the concepts of data, information and knowledge; algorithm, model and solution; big data, cloud computing, mathematical optimisation, machine learning, deep learning, language model, generative artificial intelligence and agentic system;
– explain in non-technical terms how a transformer works, with reference to tokenisation, the attention mechanism and the distinction between training and use, and describe what changes when a model moves from generating text to performing actions through external tools;
– describe the factors accounting for the pervasiveness of the quantitative paradigm — the explosion of big data, the datafication of experience and the economics of computing infrastructure — and their effects on two specific domains: information and journalism, and marketing and advertising;
– explain the current European and Italian regulatory framework governing artificial intelligence, with particular reference to transparency obligations concerning synthetic content and copyright regulation.
2.2 Applying Knowledge and Understanding
By the end of the course, students will be able to:
– design, execute, iterate and document instructions (prompts) for generative artificial intelligence systems, using at least one text-generation platform and one image-generation platform among those available at the time the course is delivered;
– produce an original piece of communication content (text, image, audio or video) using such systems and document the production process;
– apply to their own work the transparency and disclosure requirements concerning the synthetic nature of content established by the applicable regulations;
– design a multi-step workflow integrating a generative system with external sources or tools, while identifying its limitations and points of vulnerability.
2.3 Making Judgements
By the end of the course, students will be able to:
– critically assess a quantitative claim encountered in public discourse on artificial intelligence by tracing it back to its source, identifying its nature (observed data, estimate, projection or interested-party statement), and identifying any interests held by the party producing it;
– recognise and discuss forms of bias that may occur in training data and in the outputs of generative systems, with specific reference to gender representation and inequalities;
– identify errors, omissions and unverified claims in the output of a generative system, and make a reasoned decision as to whether and how it should be used.
2.4 Communication Skills
By the end of the course, students will be able to:
– explain to a non-specialist audience, using appropriate terminology and without misleading simplifications, what a generative artificial intelligence system does and does not do;
– present and defend the choices made in producing an AI-assisted communication product, accounting for the process as well as the result;
– clearly and verifiably disclose the use of generative systems in their own work.
2.5 Learning Skills
By the end of the course, students will be able to:
– independently update their knowledge in a rapidly evolving field by identifying and prioritising primary sources — providers’ technical documentation, regulatory texts and research reports with an explicitly stated methodology — over secondary publications;
– transfer the analytical method acquired during the course to tools and platforms not covered in class;
– self-assess their level of competence in using these systems and recognise the limits within which it is reasonable to rely on them.
3. Course Programme and Content
The course is divided into three parts, for a total of 40 hours: an introductory section (8 lectures, 16 hours), an in-depth section (4 lectures, 8 hours), and a laboratory section (8 sessions, 16 hours).
Part One · Basic Vocabulary (8 lectures, 16 hours)
This part establishes the shared vocabulary required to understand the subject matter of the course.
Lecture 1: Data, information, knowledge: a distinction that is not self-evident. The data–algorithm–solution triad. What it means to “make a problem computable”.
Lecture 2: A brief history of the idea of computability: algorithms and effective procedures from Euclid to al-Khwarizmi; the project of a language of calculation from Leibniz to Hilbert; the Turing machine, decidability and the Turing test.
Lecture 3: The data lifecycle: collection, storage and querying. Structured and unstructured data; an introduction to databases and database management systems; what makes big data “big”.
Lecture 4: Elements of the history of artificial intelligence: the symbolic and connectionist approaches; cycles of enthusiasm and decline; the deep-learning breakthrough.
Lecture 5: Machine-learning algorithms: supervised, unsupervised and reinforcement learning. Training, validation and generalisation. The difference between machine learning and deep learning.
Lecture 6: Language models and generative artificial intelligence: what they do, with examples from three families of tasks (classifying, answering, generating). Tokens and tokenisation; pre-training and fine-tuning; hallucinations and structural limitations.
Lecture 7: How a transformer works: the attention mechanism and encoder–decoder architecture. Multimodality: text, images, audio and video within a single model. What changes with reasoning models and inference-time computation.
Lecture 8: From chatbots to agents: tool use, access to external sources, interoperability protocols and the automation of multi-step workflows. What changes in terms of power when software acts rather than merely responds.
Part Two · In-Depth Study (4 lectures, 8 hours)
This part explores the nature of the phenomenon, its historical roots and the factors underlying its pervasiveness, with particular attention to two domains consistent with the Communication Studies curriculum.
Lecture 9: The datafication of experience and the spread of the principle of computability in economic and social life. The political economy of computation: training and inference costs, infrastructure, energy consumption and industrial concentration. Public debate on the sustainability of investment as an object of communication: who produces the figures, and with what interests.
Lecture 10: Information and journalism. How access to news is changing: platformisation, video, creators and conversational assistants. The resulting economics: from traffic to citation. AI in the newsroom: uses, safeguards and transparency policies. Trust, synthetic content and provenance verification.
Lecture 11: Marketing and advertising. From search to answers: what happens to brand visibility when the intermediary is no longer a list of links but a system that selects. Agentic commerce and interoperability standards; machine-readable product data; personalisation, measurement and new intermediaries.
Lecture 12: Knowledge, information and power. Bias and algorithmic discrimination; effects on the labour market and entry-level roles; the European and Italian regulatory framework and transparency obligations; minors and relational conversational systems. Final discussion of risks and opportunities with reference to Goals 5, 10 and 16 of the 2030 Agenda.
Part Three · Laboratory (8 sessions, 16 hours)
A series of guided exercises in designing instructions (prompt engineering) for generative models, aimed at producing original and meaningful content. Each session includes a short assignment, which is discussed collectively at the following session.
Session 1: Anatomy of a prompt: role, task, context, constraints, expected format and examples. The iteration cycle and version tracking.
Session 2: Text: generation, rewriting, summarisation, adaptation to audience and register. Systematic verification of facts and citations produced by the model.
Session 3: Images: constructing a visual prompt, controlling style and consistency, recurring limitations. Image rights, resemblance to real people and labelling.
Session 4: Audio and video: speech synthesis, adaptation and assisted editing. Practical application of disclosure requirements concerning the synthetic nature of content.
Session 5: Data: using a generative system to explore, query and visualise a small dataset; independent verification of calculations produced by the system.
Session 6: Agents and automation: building a multi-step workflow integrating external sources or tools; identifying points at which the workflow breaks down.
Session 7: Critical evaluation: group analysis of other students’ outputs to identify errors, omissions, stereotypes and unverified claims. A dedicated exercise on identifying gender and representational biases (Goals 5 and 10).
Session 8: Group project work: presentation of the communication product and the accompanying process dossier; peer discussion.
4. Teaching Methods
Teaching methods are selected according to the expected learning outcomes: interactive lectures, supported by presentations and in-class polling through Wooclap, build vocabulary and conceptual frameworks (2.1); case studies and guided exercises develop the ability to apply knowledge (2.2); critical discussion and group work develop independent judgement and communication skills (2.3 and 2.4); the process dossier supports learning and self-assessment skills (2.5).
Materials and Platforms
Presentations used in class (PowerPoint or PDF) are made available in the course section of the KIRO platform. Materials are provided in a format consistent with accessibility principles for users with disabilities, including heading structure, reading order, alternative text for images and self-descriptive links.
In-class polling conducted through Wooclap serves an exclusively formative purpose: responses are anonymous, are used to assess students’ level of understanding in real time and to inform the following lecture, and have no bearing whatsoever on the final assessment.
Attendance and Materials for Students Unable to Attend Classes
Attendance is not compulsory but is strongly recommended, particularly for the laboratory component, which is experiential in nature and difficult to replace through individual study. In any event:
– for the introductory and in-depth sections, a video recording of each lecture is available on KIRO;
– laboratory sessions are not recorded: students who are unable to attend classes take the examination according to Path B, described in Section 6, which includes an individual assignment replacing the laboratory activities.
Use of Generative Artificial Intelligence in Course Activities
The use of generative systems in preparing assignments is not only permitted but is an integral part of the learning process. Full traceability is, however, required: each assignment must be accompanied by a process dossier recording the platforms used, the main prompts, significant iterations and the checks performed on the output. Undisclosed use is regarded as a lack of professional transparency and negatively affects the assessment.
5. Reference Texts
5.1 Examination Texts (one compulsory choice for Path B)
– Alfio Quarteroni, L’intelligenza creata. L’AI e il nostro futuro, Milan, Hoepli, 1st ed., 2025, 224 pp., ISBN 9788836020089.
– Christopher Summerfield, These Strange New Minds. How AI Learned to Talk and What It Means, New York, Viking, 1st ed., 2025, 384 pp., ISBN 9780593831717 (UK edition: London, Penguin, 2025, ISBN 9781405966160).
Note on the choice. The two books are not equivalent: the first is in Italian and is 224 pages long; the second is in English and is 384 pages long and, as of the date this syllabus was prepared, is not available in Italian translation. Students are free to choose either book and are not required to notify the lecturer of their choice in advance.
5.2 Texts for Bridging Prerequisite Knowledge (recommended, not examination texts)
– David S. Moore, Statistica di base, Milan, Apogeo Education (students should use the most recent Italian edition available): introductory chapters on distributions, measures of central tendency and dispersion, correlation and inference.
5.3 Further Reading (optional)
At the end of each lecture, a specific bibliography relating to the topics covered is provided on KIRO. The following titles form the core of that bibliography and cover the course’s three main strands.
– Viktor Mayer-Schönberger, Kenneth Cukier, Big data. Una rivoluzione che trasformerà il nostro modo di vivere, Milan, Garzanti, 2013.
– Nello Cristianini, La scorciatoia. Come le macchine sono diventate intelligenti senza pensare in modo umano, Bologna, Il Mulino, 2023.
– Nello Cristianini, Machina sapiens. L’algoritmo che ci ha rubato il segreto della conoscenza, Bologna, Il Mulino, 2024.
– Kate Crawford, Né intelligente né artificiale. Il lato oscuro dell’IA, Bologna, Il Mulino, 2021.
– Cathy O’Neil, Armi di distruzione matematica. Come i big data aumentano la disuguaglianza e minacciano la democrazia, Milan, Bompiani, 2017.
– Shoshana Zuboff, Il capitalismo della sorveglianza. Il futuro dell’umanità nell’era dei nuovi poteri, Rome, Luiss University Press, 2019.
– Luciano Floridi, Etica dell’intelligenza artificiale. Sviluppi, opportunità, sfide, Milan, Raffaello Cortina, 2022.
5.4 Primary Sources and Open-Access Documents
– Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 (AI Act), with particular reference to Article 50 (transparency obligations).
– Regulation (EU) 2026/1744 (Digital Omnibus on artificial intelligence), published on 24 July 2026, amending deadlines and provisions of Regulation (EU) 2024/1689.
– Italian Law No. 132 of 23 September 2025, “Provisions and Delegations to the Government concerning Artificial Intelligence”, in force since 10 October 2025.
– Reuters Institute for the Study of Journalism, Digital News Report 2026, Oxford, 16 June 2026 (available online).
6. Assessment Methods
The examination consists of an individual oral examination lasting approximately 20–25 minutes, which may be taken in either Italian or English. Two alternative paths are available (Path A and Path B), both designed to assess all the learning outcomes specified in Section 2 of this syllabus. Students are free to choose either path; no advance notification is required, and candidates state their choice at the beginning of the examination.
6.1 Path A — Course Content and Laboratory
This path is intended for students who have attended the lectures and participated in the laboratory activities. The examination consists of three components, for a maximum overall score of 30 points:
Discussion of the product created as part of the project work and the related process dossier, submitted via KIRO at least seven days before the examination date (maximum: 9 points)
Three questions on three different topics selected by the lecturer from the introductory and in-depth sections of the course (6 points each) (maximum: 18 points)
One integrative question requiring the candidate to critically assess a case, data point or claim drawn from public debate (maximum: 3 points)
Total: 30 points
The calculation is therefore: 9 + (6 × 3 = 18) + 3 = 30 points.
6.2 Path B — Monograph and Individual Assignment
This path is intended for students who have been unable to participate in classroom activities. The examination consists of the following components, again for a maximum of 30 points:
Three questions on the monograph chosen from those listed in Section 5.1 (5 points each; maximum: 15 points)
Discussion of the individual assignment replacing the laboratory activities: three documented prompt-design exercises (one text, one image and one multi-step workflow), accompanied by a 1,500–2,000-word critical commentary and submitted via KIRO at least seven days before the examination date (maximum: 12 points)
One integrative question connecting the monograph and the individual assignment (maximum: 3 points)
Total: 30 points
The calculation is therefore: (5 × 3 = 15) + 12 + 3 = 30 points.
6.3 Assessment Criteria
Each component is assessed according to four equally weighted criteria:
Accuracy and appropriate use of terminology (accuracy of content and appropriate use of technical and disciplinary vocabulary)
Depth and ability to integrate knowledge (degree of analytical depth and ability to establish connections among knowledge acquired in different parts of the course)
Independent judgement (ability to argue a position, trace claims back to their sources and recognise the interests underlying a claim)
Communication effectiveness (clarity of presentation and appropriateness for the intended audience)
6.4 Pass Threshold, Honours and Conditions
– The examination is passed with an overall score of 18/30 or higher.
– The examination is failed, regardless of the overall score, if any of the content questions receives a score of zero or if the assignment required for the chosen path has not been submitted by the deadline.
– Honours (30 e lode) are awarded to students who achieve the maximum score and, in addition, demonstrate an outstanding level of independent judgement and communication effectiveness.
6.5 Permitted Materials and Communication of Results
– No reference materials may be consulted during the examination. The only exception is the assignment being discussed, which candidates may have in front of them while presenting it.
– The result is communicated orally at the end of the examination, together with a brief explanation specifying the score awarded for each component. The final grade is formally recorded in accordance with University procedures.
– Assignments and process dossiers are retained by the lecturer for the current academic year and remain available for consultation by students upon request.