Innovation

INESC TEC is developing the TRUST- AI project, funded by the European Union

  • Article

    Article

fib_transformacao_digital.jpg

Filipa Brito / Arquivo

INESC TECcoordinates the European project TRUST-AI, with the purpose of explaining how AI works. The specific goal is to make AI more transparent and reliable.

Artificial Intelligence (AI) or Machine Intelligence is widely used, nowadays, as regards decision making, for example, in providing personalised information or recommendation to people, namely to a doctor that is told by mathematic models where to operate, a customer that buys from one retailer rather than from another one; but it also raises questions such as the math used to equate the amount of energy consumers pay is fair and accurate.

Gonçalo Figueira, INESC TEC researcher and TRUST-AI coordinator, states that AI is increasingly influencing the way people decide, and machine reasoning is evermore abstract and complex to be easily understood. “We are talking about black boxes that find factual excellent solutions, without providing an explanation on how they do it. And that raises ethical issues, as AI is guiding human behaviour more and more every day”.

As eerie as that might seem, Gonçalo Figueira enhances that it is necessary to understand the why of each choice and that will improve the decision making process; “this project will help develop AI solutions that are more transparent, fair and explicable and, therefore, more acceptable”, Figueira affirms.

Artificial Intelligence (AI) is the branch of computer sciences that emphasizes the development of intelligence machines, thinking and working like humans. For example, speech recognition, problem-solving, learning and planning.

Regarding TRUST-AI, Gonçalo Figueira furthers, “the approach means that we put AI and humans working together to find the best solutions, that is, models that are efficient, and more easily understood, (…) by adopting an empirical learning process, human being guided and centred, which combines man-machine cognition, learning and interaction.

The resulting working tool will be applicable to three study areas, namely health, (tumour treatment), online retail purchase (to select order timetable delivery) an energy (to support demand search in building). However, this project is also applicable to other areas, such as banking, insurance, industry and public administration.

Besides INESC TEC, the project coordinator, the TRUST-AI (Transparent, Reliable and Unbiased Smart Tool for AI) relies on the participation of LTPLabs and other five partners: Tartu Ulikool (Estonia), Institut National De Recherche Eninformatique Et Automatique (France), Stichting Nederlandse Wetenschappelijk Onderzoek Instituten (Holland), Applied Industrial Technologies (Cyprus) and TAZI Bilisim Teknolojileri A.S. (Turkey).

The TRUST-AI project is financed in 4.0 million euros by the research and innovation programme of the European Union, Horizonte 2020.