A group of experts in artificial intelligence breaks with the Government due to ethical discrepancies

Relations between the Spanish government and the country's top specialists in artificial intelligence (AI) are becoming cloudy.

Last week three members of the Artificial Intelligence Advisory Council resigned, a consultative body established in 2020 to "guarantee a safe and ethical use of AI", "composed of internationally renowned Spanish experts", as described on the website from La Moncloa.

As EL PAÍS has learned, other members of the group have also considered leaving, although they have not specified it.

The trigger was the signing of a collaboration agreement between the Spanish Government and a research institute from the United Arab Emirates, ADIA Lab. The Secretary of State for Digitization and Artificial Intelligence (Sedia), under the Ministry of Economic Affairs and Digital Transformation , announced last week that the aforementioned Emirates laboratory had chosen Granada as the location for its European headquarters.

The center will receive an initial investment of five million euros, although the Emirati partners agree to increase this amount by launching various research projects.

More information

GRANDSTAND |

A decision that ignores ethical principles in artificial intelligence

The agreement "causes great concern, since it contradicts the principles of ethics and security with which the Spanish Government has committed to developing new technologies," read an open letter published this Tuesday by this newspaper, signed by four leading researchers in the discipline, two of them members until last week of the advisory council.

ADIA Lab is a scientific center "funded by a government that does not recognize the independence of science, which tramples on human rights, especially women's, LGTBQI+ and immigrant communities," recalls the text.

The ministry's decision to ally with the United Arab Emirates caused unease among a broad group of advisory council members, already upset over its token role in AI-related strategic decisions.

The reactions were not long in coming.

Some proposed a mass resignation, although that option cooled down as the hours passed.

The first to resign from the post was Carles Serra, director of the CSIC's AI Institute.

He communicated his decision on Wednesday, 24 hours after the agreement with ADIA Lab became known.

Aware that the situation required it, the ministry called an "urgent meeting" of the council the next day.

It was directed by the Secretary of State, Carme Artigas.

"Some of the members asked us to know the details of the agreement and an informal videoconference was organized," they explain from the ministry.

"Two of its members expressed their disagreement and the agreement and the impact it would have on scientific knowledge was explained to them in detail."

What was said there did not prevent two more experts from that body from leaving.

That same night Lorena Jaume-Palasí, an expert in ethics and philosophy of law applied to technology and founder of centers such as Algorithm Watch or The Ethical Society, presented her resignation.

The next day Ricardo Baeza-Yates, Director of Research at the Institute of Experimental AI at Northwestern University, in Silicon Valley (San Francisco), professor at Pompeu Fabra University (Barcelona) and National Computing Award winner, did so.

None of the three experts has received any communication from the Secretary of State after presenting his resignation.

“Due to the heterogeneity of its members, it is difficult to find consensual positions, which enriches the debate and provides new points of view.

Therefore, the disparity of criteria between members is common, always with respect for all opinions ”, argue sources from the Secretary of State.

“We consider it very positive that an institution of scientific prestige that has Nobel laureates on its board chooses to settle in Spain rather than in other European countries such as France or Germany.

We respect the decision of the members of the advisory board who consider that they no longer want to continue contributing their vision to this body and we appreciate their contribution to date", they add from Sedia.

“A government agreement to install an AI laboratory in Spain financed by an autocratic country that violates human rights, particularly those of women, is not ethically acceptable,” explains Baeza-Yates.

“In fact, it is incompatible with the very spirit of the advisory council, which has as one of its missions to ensure the ethics of AI in Spain.

For these reasons I cannot legitimize with my silence initiatives that violate basic principles of scientific ethics and therefore I resigned from the council”.

"The decision to sign an agreement with ADIA Lab has not only been a very serious straw that broke the camel's back: it is an indicator of the defective dynamics of the advisory council and of a position on the part of the Secretary of State that is in deep contradiction with the ethics and security objectives for which the advisory council was created”, points out Jaume-Palasí.

As this newspaper has learned, a group of NGOs focused on human rights, digital rights and immigration are also preparing an open letter in which they make clear their rejection of the agreement signed between the Government and the research center of the United Arab Emirates.

Another recurring complaint among some members of the group of experts is that the Government did not go to them for advice before making strategic decisions.

“The council has mainly had a decorative role.

We have not been consulted to assess the impact of AI systems or issue recommendations on really relevant issues”, complains the philosopher.

"For two years now, the few meetings that we have held have been more of a declarative nature in which we were informed with very condensed presentations about the projects of the Secretary of State."

The signatories to the letter, including Ramon López de Mántaras, one of the pioneers of AI in Europe, and Carmela Troncoso, an expert in algorithmic security and developer of the technical protocol used in covid

tracking

, recognize the courage that the Government had in identifying the challenges that AI poses for society and in wanting to take action on the matter.

The advisory council was created in 2020 to "provide independent recommendations on steps to take to ensure safe and ethical use of AI."

But they regret that, by allying with the United Arab Emirates, "ethical principles in the economy, in science in general and in AI in particular" are ignored.

Troubled waters at the Secretary of State

The departure of 3 of the 18 experts that make up the advisory council is not the first conflict that has arisen in the Secretary of State.

The department has experienced several convulsions since it came to light in January 2020. The head of the department, Carme Artigas, has had three chiefs of staff in less than three years.

Several advisers have left the team in this time, including signings made at the express request of Artigas, as is the case of data scientist Mara Balestrini, who left the position three months after arriving.

There were also casualties among high-level officials, who sought accommodation in other ministries, as published by

.

Among them Fernando de Pablo, Secretary General until May 12, 2020;

María José Gómez, Deputy Director General of Talent and Digital Entrepreneurship;

David Pérez, coordinator of the Technology and Digital Entrepreneurship area, or Amparo Peris, deputy general director for the Digital Society.

Two months after the department led by Artigas was established, Spain was confined.

The Secretary of State led the project to develop a covid case tracking application, which involved the coordination of various ministries and working groups.

Most of the departures from Artigas' team occurred during that process.

The application was only able to notify 150,000 infections in almost two years.

You can follow

Subscribe to continue reading

Read without limits

Keep reading

I'm already a subscriber