Science and Technology

The future of artificial intelligence discussed at Técnico

Técnico and the Público newspaper organised a debate on the impact of these tools in Academia and teaching methods.

Artificial intelligence (AI) was the topic of a debate organised by Instituto Superior Técnico in partnership with PSuperior (Público newspaper) on May 8th, at Técnico Innovation Center powered by Fidelidade, as part of the JobShop organised by the Técnico Students’ Union (AEIST). The debate, which was also live-streamed, brought together dozens of spectators who attended the debate “How will Artificial Intelligence revolutionise Academia and teaching methods?”.

The speakers panel included Mário Figueiredo, a professor at Técnico and a researcher at Instituto de Telecomunicações (IT), Sancha Barroso, a Computer Science and Engineering MSc student, José Varela, AI Evangelist at NTT Data, and Paulo Dimas, CEO of the Centre for Responsible AI and Vice-President of Innovation at Unbabel, a Técnico spin-off. The debate was moderated by Victor Ferreira, a Público newspaper journalist.

‘It’s a great pleasure to host this initiative here’, said Rogério Colaço, president of Técnico, stressing that this is not the first time that the Técnico has hosted a debate in collaboration with Público. David Pontes, the newspaper’s director, thanked Técnico for hosting the event at the ‘fantastic facilities’ of Técnico Innovation Center and argued that IA ‘still has a lot to give’.

The conversation began with a question to Sancha Barroso about how AI has impacted her student experience over the last few years. “It was definitely a turning point”, she explained, referring particularly to productivity issues. While writing her master’s thesis, she has seen tools like ChatGPT as ‘personalised learning support’, stressing that she doesn’t use it as a first resort.

‘In manual tasks that don’t require complex reasoning, but which take time,’ she says that by using AI, ‘she can do it in a matter of seconds’ and can concentrate her thoughts ‘on the more fundamental objectives’ of her work. As an example of this type of task, she mentioned formatting changes of a large number of tables in a document.

‘The machine has already replaced the student. When will it replace the professor?’ – the moderator posed this challenge to Mário Figueiredo, but the professor promptly refuted it, laughing. ‘The machine at most replaces the student in assessment, but it doesn’t replace them in learning,’ he defended. ‘When ChatGPT appeared, people were worried – ’now how are we going to assess students? According to the Técnico professor, this reaction ‘says more about assessment than it does about ChatGPT’.

In his opinion ‘an excessive focus on assessment’ is reflected in both professors and students. The researcher sees AI tools as a disruptive element ‘in a good way’ which, however, ‘doesn’t replace the more human or relational part of stimulating curiosity and appreciation for subjects’. The professor also emphasised that he ‘doesn’t teach anyone anything’. Instead, his role is to ‘create in other people the desire to learn’. ‘This is a good time to rethink the foundations of education,’ he emphasised.

Although the debate focussed on how AI impact academia, it wasn’t just academics who took the floor. At the event – which welcomed questions from viewers of Público’s livestream – there was also room for business representatives share their perspectives. Paulo Dimas recalled that ‘there’s talk of productivity increases of 20 to 30 per cent in terms of code production’ when some AI tools are used, which “is great because it saves hours of development time”. He also emphasised the usefulness of these mechanisms for simplifying and explaining large amounts of code to a user quickly and clearly.

José Varela sees it as ‘absolutely basic and fundamental to know how these models work, experimenting and realising in which cases they work well and in which cases they work badly’. He gave the audience the example of a hypothetical person who knows how to mentally solve complex multiplications, in which case, he said, he/she would be expected to know how to do simpler operations (such as short sums). ‘This is not true for artificial intelligence models,’ he warned – ‘they do some things brilliantly, while they have extreme difficulty doing other things that would be almost basic for us.’

The debate can be seen in full on the Público website.