Features

AI explosion

Barcelona has joined the European Laboratory for Learning and Intelligent Systems (ELLIS), a network for excellence in AI research

“It’s very important that the people developing these technologies are aware of the benefits and risks they entail”

In 2019, pe­cu­liar mon­i­tor­ing tests for school­child­ren began in China. Using an ad­vanced com­puter vi­sion and ar­ti­fi­cial in­tel­li­gence (AI) sys­tem, the chil­dren’s bio­met­ric data, such as eye move­ments or fa­cial ex­pres­sions, were recorded and analysed to see if they were pay­ing at­ten­tion and con­cen­trat­ing in class. AI had en­tered the class­room, sup­pos­edly to con­tribute to im­prov­ing the pupils’ at­ten­tion and pro­mot­ing bet­ter aca­d­e­mic per­for­mance.

This very con­tro­ver­sial ex­per­i­ment would have no place in our ed­u­ca­tion sys­tem, but AI is find­ing its way into class­rooms in other ways. Teach­ers from a pri­vate school in Ter­rassa have taken part in a train­ing pro­gramme in which AI and vir­tual re­al­ity are used to im­prove the com­mu­ni­ca­tion skills of teach­ers. Ac­cord­ing to sources at the Gresol In­ter­na­tional Amer­i­can School, the AI helps to “im­prove the wave­length of the voice, to con­trol the char­ac­ter­is­tics of the speech and to prac­tise breath­ing and nerve con­trol”.

This is just one ex­am­ple of how AI is gain­ing a greater pres­ence in dif­fer­ent as­pects of our lives. One area where it has great po­ten­tial for ap­pli­ca­tion is in health care. The Parc Taulí hos­pi­tal in Sabadell, for ex­am­ple, has cre­ated an emer­gency unit made up of a team of nurses sup­ported by soft­ware, called Medik­tor, which is based on ar­ti­fi­cial in­tel­li­gence. The soft­ware guides them and helps them treat pa­tients with low-com­plex­ity and high-in­ci­dence patholo­gies, such as cer­vi­cal pain, sprained an­kles, uri­nary in­fec­tions or ver­tigo. The hos­pi­tal claims this “al­lows wait­ing times to be sig­nif­i­cantly re­duced with­out sac­ri­fic­ing the qual­ity and safety of the care”.

The im­pact of gen­er­a­tive AI tech­nol­ogy, pop­u­larised by Chat­GPT, can go still much fur­ther. An ex­am­ple is the in­te­gra­tion of the tech­nol­ogy into in­ter­net browsers, as is the case with the Opera browser. An­other ex­am­ple was seen in Barcelona on June 14, when the city hosted the first film fes­ti­val in Eu­rope fea­tur­ing films gen­er­ated by AI. The event, called +Rain Fest, was or­gan­ised by the Pom­peu Fabra Uni­ver­sity (UPF).

In fact, the UPF is one of the 10 Cata­lan cen­tres that are car­ry­ing out ad­vanced re­search in the field of ar­ti­fi­cial in­tel­li­gence and con­tribut­ing tal­ent to the new Eu­ro­pean Lab­o­ra­tory for Learn­ing and In­tel­li­gent Sys­tems (ELLIS). The Ellis net­work is a Eu­ro­pean ini­tia­tive to bring to­gether and co­or­di­nate the best re­searchers in the field and en­sure last­ing in­ter­na­tional lead­er­ship of AI made in Eu­rope. The idea, which came from the sci­en­tists them­selves, is to have a multi-cen­tre re­search lab­o­ra­tory, which al­ready brings to­gether top-level in­sti­tu­tions in 14 coun­tries, to push the sci­en­tific and tech­no­log­i­cal lim­its of a safe AI for so­ci­ety.

“AI in Eu­rope takes a more hu­man­is­tic ap­proach, giv­ing im­por­tance to human rights, while in the United States, due to cor­po­rate in­flu­ence, they focus more on per­for­mance. And in China data pro­cess­ing has com­pletely dif­fer­ent rules of the game. In Barcelona, we have a pow­er­ful group aimed at con­tribut­ing to es­tab­lish­ing the foun­da­tions of this third way and help en­sure that Eu­rope does not miss the AI train,” ex­plains Di­mos­the­nis Karatzas, co-di­rec­tor of the Barcelona Ellis unit to­gether with Carme Tor­ras, who adds: “You need to in­sist a lot on train­ing. It’s very im­por­tant that the peo­ple de­vel­op­ing these tech­nolo­gies are aware of the ben­e­fits and the risks they en­tail.”

The Cata­lan Ellis group is made up of 21 re­searchers from five uni­ver­si­ties (Uni­ver­sity of Barcelona-UB, Au­tonomous Uni­ver­sity of Barcelona-UAB, Poly­tech­nic Uni­ver­sity of Cat­alo­nia-UPC, Open Uni­ver­sity of Cat­alo­nia-UOC and UPF) and five re­search cen­tres: the Com­puter Vi­sion Cen­ter, the In­dus­trial Ro­bot­ics and In­for­mat­ics In­sti­tute, the Barcelona Su­per­com­put­ing Cen­ter, the Ar­ti­fi­cial In­tel­li­gence Re­search In­sti­tute and the In­tel­li­gent Data Sci­ence and Ar­ti­fi­cial In­tel­li­gence Re­search Cen­ter. Al­though new, it is one of the Eu­ro­pean net­work’s largest units and has room to con­tinue grow­ing.

“We al­ready have a crit­i­cal mass in AI re­search, but split among sev­eral in­sti­tu­tions and re­search groups and not nec­es­sar­ily aligned, so this unit will allow us to struc­ture re­search and pro­mote co­op­er­a­tion,” says Karatzas. The Com­puter Vi­sion Cen­ter’s as­so­ci­ate di­rec­tor also hopes that the ini­tia­tive will help re­tain tal­ent, which he says often moves out­side Eu­rope.

Mean­while, the Cata­lan busi­ness min­is­ter, Roger Tor­rent, says that Barcelona’s in­te­gra­tion into the Eu­ro­pean AI net­work “po­si­tions Cat­alo­nia as a point of ref­er­ence in Eu­rope in terms of re­search and tech­no­log­i­cal ex­cel­lence, and helps to gen­er­ate and at­tract new re­search tal­ent.”

fea­ture Tech­nol­ogy

Dimosthenis Karatzas

“It is being rolled out without assurances”

The AI boom is happening now but research has been going on in Barcelona for some time.
The first doctoral thesis in AI was done here 40 years ago, by Ramón López de Mántaras, the founder and director of the Artificial Intelligence Research Institute of the CSIC. Since then, the ecosystem has grown steadily and evolved into various verticals specialising in robotics in computer vision, in ethics... We haven’t popped up like a mushroom, we’ve been working on this for many years and have covered everything, from basic research to the economic and social impact.
How are you experiencing this AI boom that came with ChatGPT?
The main problem is that ChatGPT is what we call a language model, which is trained with unknown knowledge to synthesise natural language. The idea is that it generates a language that sounds good, but does not have to be correct.
And in some cases it’s given too much credibility?
Exactly. The problem is that the industry is testing it live. Developers say that the system has bugs and that the best way to find them is to take it onto the street and see what goes wrong. It’s an unusual policy. Can you imagine if car manufacturers did the same?
Can the eagerness of the industry be dangerous?
When private interests are mixed with something that is emerging and explodes onto the scene this way, it can end up being a dangerous mix. AI by itself is a discipline that can help in many things and have a very important effect on society, but if we launch it without control and the commercial interests of companies are prioritised, we will have a problem. And we will see that sooner rather than later.
In what sense?
We already have a dual problem. There’s a lack of knowledge about these systems, about what they can and can’t do, and there’s a desire to integrate them into real products, for the benefits of these companies, but without enough time having passed to create a legal context to regulate them.
And this will take time because the authorities work slowly.
Of course, it will take time. That’s why the initiatives of some countries like Italy, which have banned its use, can provide time to manage it. Right now it’s being rolled out quickly, with little oversight and no assurances. It has been launched without having done the necessary testing or properly explaining what it is for.
Sign in. Sign in if you are already a verified reader. I want to become verified reader. To leave comments on the website you must be a verified reader.
Note: To leave comments on the website you must be a verified reader and accept the conditions of use.