There are no specific prerequisites for this course. However, it is recommended that students have a basic understanding of programming and computer science, which can be acquired by taking the “Computer Science for Cultural Heritage” course in the first semester or another introductory computer science course. During the course, certain concepts from basic calculus will be covered in a simplified manner (e.g., sets, functions, and the minimum and maximum values of a function)

The course provides an overview of the main concepts related to AI and its models.

Knowledge and Understanding

Upon successful completion of the course, students will have acquired a comprehensive understanding of the fundamental concepts of Artificial Intelligence (AI), its main subfields, and the relationship between learning, intelligence, and decision-making processes.

Students will understand the general functioning of modern generative AI systems for text and image creation, the principles of prompt engineering, and methodologies for designing structured prompts tailored to specific objectives.

They will acquire the mathematical foundations required to understand Machine Learning models, including the concepts of functions, dependent and independent variables, graphical representations of functions, and linear models.

Students will understand the theoretical foundations of Machine Learning, distinguishing between supervised, unsupervised, and reinforcement learning approaches, as well as the main techniques used for regression, classification, and clustering problems. They will also gain knowledge of artificial neural networks, deep learning, and convolutional neural networks for image processing and classification.

Furthermore, students will acquire knowledge of AI applications in the cultural heritage domain and develop an understanding of the major ethical, social, legal, and environmental challenges associated with the development and deployment of AI systems, including algorithmic bias, hallucinations, misinformation, copyright issues, deepfakes, and the European regulatory framework established by the AI Act.

 

2. Ability to Apply Knowledge and Understanding

Upon successful completion of the course, students will be able to:

• Use generative Artificial Intelligence tools to support content creation, information retrieval, data analysis, and problem-solving activities.
• Design effective and structured prompts, adapting them to different contexts and desired outcomes.
• Analyze simple datasets and interpret relationships between variables using linear regression and classification models.
• Use conversational AI tools for data visualization and for supporting the analysis and interpretation of Machine Learning models.
• Apply basic supervised and unsupervised learning techniques to solve prediction, classification, and clustering problems.
• Perform introductory clustering and data analysis exercises using dedicated development environments.
• Interpret the general functioning of neural networks and deep learning models in image classification tasks.
• Identify and discuss AI use cases in cultural heritage and other professional domains.

3.Independent Judgment

Upon successful completion of the course, students will be able to critically evaluate the opportunities and limitations of Artificial Intelligence systems, distinguishing between statistical correlation, causation, and evidence-based conclusions.

Students will be capable of interpreting the outputs of machine learning models, identifying potential sources of error, uncertainty, or bias, and assessing the reliability of information generated by AI systems.

They will also be able to formulate informed and independent judgments regarding the social, ethical, legal, and environmental implications of AI technologies, with particular attention to transparency, fairness, intellectual property rights, and the protection of fundamental rights.

 

4.Communication Skills

Upon successful completion of the course, students will be able to communicate concepts, methods, and applications of Artificial Intelligence using appropriate technical terminology and language suitable for both academic and professional contexts.

Students will be able to present and interpret results derived from data analysis and machine learning models, clearly explaining methodological choices and critically discussing the advantages, limitations, and implications of the adopted approaches.

They will also be able to effectively communicate practical and project-based outcomes through graphical representations, data visualizations, and digital presentation tools.

 

5.Learning Outcomes

Upon successful completion of the course, students will have developed the methodological and conceptual tools necessary for the continuous updating of their knowledge and skills in a rapidly evolving technological and scientific field.

Students will be able to independently consult technical documentation, scientific literature, and specialized sources related to Artificial Intelligence, critically assessing their quality and reliability. They will also be capable of autonomously learning and evaluating new AI tools, models, and technologies, adapting them to different study, research, and professional contexts, particularly within the fields of Digital Humanities, Cultural Heritage, and digital transformation.

 

The course mixes traditional lectures with hands-on workshops. At the beginning of the semester, the students are assigned a project involving the use of conversational AI that will be presented during the final lesson.

Oral exam about theoretical concepts.

To take the oral exam, students must have submitted the project assigned during the course and the project has to be evaluated positively.

 

The oral exam will be based on the study of the teaching material provided on the teams channel of the course. The oral exam is to verify the knowledge of the lesson topics, analyzed throughout the course. In particular, the student will be evaluated considering:

• knowledge and understanding of the subject matter,
• ability to apply knowledge acquired, synthesis and formal correctness in the exposition and ability to argue one's thesis
• autonomy of judgment,
• communication skills.

 

The exam score is assigned using a grade expressed in thirtieths (or through a pass/fail assessment).

 

In the evaluation of the exam, the determination of the final grade considers the following elements:

 

28–30 with honors: excellent evaluation; comprehensive and in-depth preparation; in the presentation, precise expression, confident mastery of the topics, and clear explanation of the definitions required for the scores in the previous ranges, as well as: the ability to explain the main components of a neural network, examples of layers in a convolutional neural network, and the use of the gradient descent method in the context of the models covered in the course. Explanation of the concept of weights in a neural network.

27–25: Knowledge and correct explanation of the definitions required for scores in the 20–18 and 24–21 ranges, as well as examples of problems to which neural networks and deep neural networks can be applied. Knowledge of the equations of the generic hypotheses for linear and logistic regression.

24–21: Knowledge and correct explanation of the definitions required for the 20–18 score range, as well as examples of problems that can be solved using linear regression, logistic regression, and classification models.

20-18: knowledge and correct exposition of the following definitions:

Artificial Intelligence System Definition. Artificial intelligence and its sub-disciplines. General Purpose AI. General-purpose AI model v/s AI system. Generative AI. Large Language model. Human Intelligence characteristics and definition. Learning. Deductive vs inductive reasoning/learning.

Machine Learning. Supervised Learning. Regression models. Classification models. Tree-based structures. KNN. Unsupervised learning. Clustering. Reinforcement learning. Outliers. Neural Networks. Deep Learning. Convolutional neural networks. AI hallucinations.

The student must be able to present a coherent discussion on the social and ethical AI challenges.

 

 < 18: insufficient evaluation; insufficient preparation; insufficient presentation.

 

 

Assessment sessions are scheduled through the EasyTest platform, in

accordance with the number of sessions and time windows set by the

Teaching Calendar approved by the Department Board.

Examination board: Prof. Carla Di Biccari (Chair); additional members will

be appointed according to Department procedures.

Office hours: by appointment, to be arranged by e-mail.

Teaching materials (slides, exercises, examples, and project instructions) will be made available through the institutional channels indicated by the lecturer at the beginning of the course. …

- Definition of Artificial Intelligence. Sub-areas of Artificial Intelligence. The link between Intelligence and learning. Main human and mammals learning mechanism.

Current most used conversational AI interfaces and underlying models for text creation.

-Current most used conversational AI interfaces and underlying models for image creation. Prompt engineering. Creation of a structured prompt using personas.

-Recall of mathematical concepts useful to understand the basics of Machine learning: Concept of function, set, domain and range/co-domain. Math functions in R2 , Graph of a function, Function notation, maximum and minimum of a function. General equation of a line in R2.

-Introduction to Machine learning. Definition of Machine Learning, the relationship between statistics and machine learning in relation to their different objectives and the use of the same models. Causality and correlation in statistics. Dependent and independent variables. The regression models. Linear regression with two variables. Supervised learning and the linear regression as a supervised ML model. Labeled data. Using the Gradient Descent algorithm to find the parameters of the linear regression model. Using conversational AI to plot a dataset, finding parameters for linear regression given a dataset.

- Machine learning. Supervised learning: classification. Types of classification. Supervised classification algorithms/models: Logistic regression, K-nearest-neighbors, classification/decision trees (CART).

Using conversational AI to plot a dataset, finding parameters for logistic regression given a dataset.

- Machine learning. Unsupervised learning: k-means clustering. Exercise on google colab.

Reinforcement learning. Machine learning steps. How to choose a model.

-Neural networks and deep learning. The components of a neural network. Deep learning for image classification. How a computer stores pictures: tensors. Convolutional Neural networks.

-AI use cases for cultural heritage.

-AI social and ethical challenges. Hallucinations. Algorithmic bias and discrimination. Copyright issues. Deep fakes, misinformation. Energetic issues. The European AI ACT.

Teaching material (slides, datasets, useful links) is made available on the teams channel of the course.

Some topics are covered by the book:
Machine Learning For Absolute Beginners: A Plain English Introduction (Learn AI & Python for Beginners). Oliver Theobald. (2018)