The company EXPAI is working towards achieving more understandable, safer and fairer artificial intelligence, and its methodology enables detecting racist, sexist or LGBTI-phobic algorithms. Its founder partners are young teachers and students of the University in the fields of ICT and entrepreneurship.
The entrepreneurial team consists of four partners: Javier Rando , a final year student of the bachelor’s degree in Mathematical Engineering in Data Science (UPF); Pablo Vitoria , student of the final year of the bachelor’s degree in International Business Economics (UPF); Daniel Hormigo and Carlos Soler , associate professors of the UPF School of Engineering.
Transparent AI at the service of companiesArtificial intelligence is being increasingly used, but there are a few techniques to ensure that it is a fair and safe tool. EXPAI offers companies a solution that allows understanding how their artificial intelligence models work without the need for programming. It is a visual tool that can be used by anyone within a company regardless of their technical knowledge.
EXPAI will allow algorithms to cease to be opaque and for anyone to be able to understand how they are taking decisions
"We saw the opportunity to create a solution that allows companies to assess and understand their machine learning models intuitively and visually. EXPAI will allow algorithms to cease to be opaque and for anyone to be able to understand how they are taking decisions. This transparency will improve the adoption of artificial intelligence by companies because it facilitates an understanding of the process and helps to make better decisions", explains Rando , co-CEO and head of ICT at EXPAI.
Algorithms are "black boxes"Today, companies place contracting processes, the granting of loans, and sales strategy in the hands of artificial intelligence. The danger of leaving decisions to these algorithms is their lack of transparency. "It is often said that algorithms are black boxes: you know what goes in, you know what comes out, but you never know what happens inside. At EXPAI, we have developed a methodology to explain what happens inside and to know in detail the reasons that have led the algorithm to take each of its decisions", Rando points out.
EXPAI’s methodology enables companies and administrations to verify that their mathematical models work correctly or incorrectly and identify errors or biases, but also the potential to explore algorithms to the fullest.
A startup in the process of validating the market ideaThe team of entrepreneurs has enjoyed the support of the UPF Entrepreneurship Service, which has helped them get access to resources for statups such as 4YFN , and the UPF Hacklab. "Right now, we are in the process of validating the market idea, with very good results, so we hope to get our first customers shortly. Thanks to UPF we will be at the 4YFN, and we are convinced it will be a great platform to give us visibility", the EXPAI partners assert.