ChatGPT per tutti, artificial intelligence in the East Campus

e6a7dcdb2f51a1f28cc88ecf27f63c37.jpg
b14c19cf85ee28859edebf30e9faca31.jpg
2ce695a8967406efe0dafca9f6207c40.jpg
1f29d04fcc25dd35b7d37e2c45f19e91.jpg
843e9f41fb1afe11d8786a8668de965a.jpg
f51c1533c9da7a905b09210587588016.jpg
fc260a1250c7c08faa39a61d90395da4.jpg

Institutional Communication Service

3 April 2023

The Università della Svizzera italiana established itself as one of the places to discuss Artificial Intelligence on Friday afternoon, combining cutting-edge academic research with a focus on ethical and social issues and communication. The ChatGPT per tutti meeting, in the multi-purpose hall of the East Campus Lugano, was attended by a large audience of all ages. Presenting the popular application developed by OpenAI were Professor Luca Maria Gambardella, a leading international expert in artificial intelligence, and Andrea Scarinci of Lugano Living Lab.

Following an introduction on the development of artificial intelligence and the various technologies available, Professor Gambardella focused on deep learning, a system also used by ChatGPT: an artificial neural network that learns from large amounts of data how to perform certain tasks, such as writing texts or answering questions. Sometimes it occurs that the training data contains prejudices or discrimination that are then incorporated into the system but, as Professor Gambardella stated, artificial intelligence does not create these prejudices but, on the contrary, helps to highlight them.

At the Università della Svizzera italiana, these topics are studied in depth not only from a technological point of view, but also with regard to ethical, social and philosophical implications. Professor Gambardella concluded his talk by pointing out two possible risks: on the one hand, laziness in delegating more and more tasks to ChatGPT and similar systems; on the other hand, the loss of critical sense towards information. Risks that were also at the centre of Andrea Scarinci's demonstrations that showed how relatively easy it is to have ChatGPT write texts containing false and absurd information, also bypassing the filters introduced by the developers.

 

Faculties

Sections