Professor Nicola Mazzari tells us about how algorithms and AI can be applied and used in education.

When talking about artificial intelligence applied to teaching, one might think of the stereotype of using ChatGPT and similar to “cheat” in exercises. The topic, in reality, is much broader and indeed, in the classroom you yourself recently demonstrated that AI can be used for teaching purposes..

ChatGPT is an excellent tutor, “someone” to question on specific topics and with whom you can have a dialogue that allows you to delve deeper into unclear points. This works very well for notional topics well established in the literature. So it also works well with mathematics as long as you are asked to give examples, general explanations and the like. However, LLM algorithms are not designed to do calculations and logical reasoning even if they can still help in setting up standard exercises. So I told the students to use it from day one because I can only respond to them one at a time, while ChatGPT can interact individually with everyone at the same time. This allows students to do experiments and get help. Maybe you get the right answer, but you start an almost dialectical process to think about. Then the teacher, once this “gymnastics” has been completed, can intervene to clarify any remaining doubts.

In addition to ChatGPT I told the students to use, a very reliable calculation service available online. In this way I could give exercises that were not easy and I gave the students powerful tools to solve them. I believe this process facilitates access to this subject and the curiosity to learn the methods to arrive at solutions.

As for the exam of the “Mathematics for Business” course, part of this is always a report. Now, on any topic ChatGPT can write a reasonable report and so this year I asked the students exactly this: “let ChatGPT work and then tell me if you learned something”. In this way, instead of reading in books, web pages, etc. the information to summarize it and perhaps do some exercises, I had them do an “experiment” and the report consisted of drawing conclusions. One of the aims of the course is to learn a scientific method and the tools of algebra and mathematical analysis. So having them do an experiment – which depends among other things on how they individually interact with the bot – is a good way to get different results: individual reports showing different characteristics due to the mode of interaction; habit of experimenting and using modern tools; answers that depend on the personal capacity for in-depth analysis.

Furthermore, in this way students immediately learn the possible limits of these technologies and therefore an informed way in their use.

H-FARM College for the 2024/25 academic year has decided to focus on artificial intelligence by adding two “tech” courses to the economic study ones. What is the added value in terms of skills and theoretical notions that these two paths can offer?

Machine Learning techniques are based on the development of mathematics, statistics and computer science.

These subjects can be studied in many degree courses almost everywhere in the world and for several years. What has changed in recent years is the accessibility of these tools. Therefore, in addition to the development of AI, based on machine learning, we need people capable of understanding and using the tools now available for example in the economic sector.

The artificial intelligence we have now does not do intelligent things, but it allows us to scale and bring value to a whole series of fairly elementary activities. It is therefore urgent to train students capable of understanding the development of these technologies and of using them immediately to guide future scenarios.

The educational use of artificial intelligence is a topic that not only concerns students and the education sector, but which involves us all as a society. Enthusiastic approaches on the one hand, prejudices and fears on the other… From this point of view, what is the situation in Italy in your opinion? How can we encourage a constructive dialogue on the potential (and limits) of these new technologies?

This question is very difficult…

The thing I like about this technology is that it does “stupid” things and therefore it can be used by everyone. When this is the case, the impact is strong.

Some time ago there was hype about blockchain technology, which is certainly not easy to understand and use. So it has an impact, but it remains something foreign to most people. On the contrary, using ChatGPT, or other LLMs, is immediate and allows you to do simple things. So I think the impact will be notable, in the sense that it is a technology that will easily enter everyone’s daily life.

There are also many concerns about the use that will be made of it, I am thinking for example of the issue of information and propaganda. However, this is a topic that I have not yet explored.

Read the interview with Professor Mazzari on StartupItalia