MEPs call for greater regulation of AI discrimination

"We have fought for decades to assert our values of inclusion, non-discrimination, multilingualism, and cultural diversity which our citizens consider essential to our European identity. These values must also be reflected online", said the opinion's rapporteur, Sabine Verheyen (EPP), after the vote. [BEST-BACKGROUNDS/Shuttestock]

The European Parliament’s culture and education committee endorsed on 15 March an opinion calling for a framework to regulate artificial intelligence (AI) in order to “reduce gender, social or cultural bias in technology”, and the Commission is expected to propose a legislative framework on the matter in April. EURACTIV France reports.

Concerned about the impact that the use of artificial intelligence technologies can have on “the backbone of our society’s fundamental values and rights”, the European Parliament’s culture and education committee called for increased regulation in the text.

In its opinion, adopted with 25 votes in favour and four abstentions, lawmakers stressed that “to be ethical, AI must be developed, deployed and used in a sustainable and socially responsible manner, following a gender equality strategy, respecting cultural diversity, promoting digital literacy, closing the digital gap and safeguarding intellectual property rights.”

The group also noted that “the crucial importance of a coherent vision and strategy at Union level in order to achieve a genuine digital single market within an AI-powered society that would fully benefit users.”

While MEPs said they want to fight against the use of biased data that could reproduce or deepen existing gender inequalities or societal discrimination, MEPs also called for a “clear ethical framework” to ensure cultural and linguistic diversity in culture and media – by regulating algorithm-based content recommendations for streaming services for example – to preserve European diversity.

“We have fought for decades to assert our values of inclusion, non-discrimination, multilingualism, and cultural diversity which our citizens consider essential to our European identity. These values must also be reflected online”, said the opinion’s rapporteur, Sabine Verheyen (EPP), after the vote.

The opinion, which has yet to be adopted in the plenary, follows the EU AI White Paper which called for a “European approach of excellence and trust”.

The Commission is expected to propose a legislative framework on the subject in April.

EU must 'proactively' tackle AI discrimination, Jourová says

The EU should not ‘copy and paste’ everyday racial discrimination and bias into algorithms in artificial intelligence, the EU’s Vice-President for Values and Transparency Věra Jourová has said.

An “inclusive AI” label

“Beyond the discrimination aspects, it is a model of society that is at stake,” Cristina Lunghi, president and founder of Arborus – an endowment fund that works for gender equality, particularly in the business sector, through the awarding of several labels – told EURACTIV France. “On a global scale, [the lack of diversity] creates economic, societal, and climate imbalances,” she added.

Last September, the Arborus team launched the new GEEIS IA label, specifically dedicated to the use of artificial intelligence in the workplace.

“We were already thinking about it, about digitalisation, about the digitalisation of jobs, about the digital transition” before the health crisis, said Lunghi, adding that because of “AI’s very strong acceleration during the pandemic”, we thought “there was a real issue on IT tools.”

“We realised that the algorithm is only a reflection of human biases. Since these biases are extremely masculine, sexist, and discriminatory, in the end, the algorithm only amplified this phenomenon”, she stressed.

Prior to the label’s launch, Arborus had inaugurated the International Charter for an Inclusive AI on 21 April of last year. This charter commits the signatory companies, including Orange, Danone, EDF, and L’Oréal, to “promote gender diversity in teams working on AI-based solutions”, to “raise awareness and make designers, developers and all those involved in the AI industry aware of stereotypes and biases that can lead to discrimination”, while ensuring “the quality of the data used to guarantee the fairest possible systems”.

While there is obviously work to be done on the technical nature of AI-powered tools, Arborus is betting on raising awareness and training the players involved in the design and implementation, but also on diversity within the very teams working on the products. “We realise that in Europe we don’t even have 20% women in the AI sectors, which are the jobs of the future,” said Lunghi.

Orange, a founding member of the Arborus Foundation, will be the first group to receive the label next month. The labels are awarded for a period of four years, with an intermediate audit after two years conducted by Bureau Veritas.

MEPs mull over 'significant risks' of Artificial Intelligence

MEPs have been urged to consider the ‘significant risks’ that next-generation Artificial Intelligence applications could entail, particularly with regards to discrimination, employment, and social exclusion.

Edited by Samuel Stolton

Subscribe to our newsletters

Subscribe