EU “discrimination by AI is a bigger concern than the extinction of humanity”

The EU’s competition chief said discrimination due to advances in artificial intelligence was a more pressing concern than the extinction of humanity.

Margrethe Vestager told the BBC that “guardrails” are needed to counter technology’s biggest risks.

He said this is key when AI is used to help make decisions that can affect people’s lives, such as whether they can apply for a mortgage.

The European Parliament is set to vote on the proposed AI regulation on Wednesday.

The AI ​​law is being considered by politicians amid warnings against the premature development of technology that allows computers to perform tasks that normally require human intelligence.

But Vestager said the potential for AI to amplify prejudice and discrimination, which could be included in the vast amount of data that comes from the internet and is used to train models and tools, is a more pressing concern. said.

“probably [the risk of extinction] It might be, but I think it’s highly unlikely. I think the risk of AI is greater than people being discriminated against [against]they are no longer seen for what they are.

“If it’s used by banks to decide whether to take out a mortgage, or your municipality’s social services, you want to make sure you’re not discriminated against. [against] It’s because of your gender, skin color and zip code,” she said.

Ireland’s data protection authority announced on Tuesday that it had put Google’s planned EU deployment of its AI chatbot, Bird, on hold.

The company was informed by Google that a ChatGPT competitor will be introduced to the EU this week, but has yet to receive details or information on how the company has identified and minimized data protection risks to prospective users. said no.

Deputy Commissioner Graham Doyle said the DPC wanted the information “as a matter of urgency” and had raised a further data protection inquiry regarding this information with Google.

In an exclusive interview with the BBC, European Commission Vice-President Vestager said AI regulation needs to be a “global issue”.

But ahead of the European Parliament’s vote on the AI ​​law, he argued that getting an agreement between “like-minded” countries should be a priority before gaining the participation of further jurisdictions such as China. bottom.

“Let’s start working on the UN approach, but we shouldn’t hold our breath,” she said.

“We should do what we can here and now.”

Mr. Vestager is spearheading an EU effort to jointly develop a voluntary code of conduct with the US government. Companies that use or develop AI will sign a set of non-binding standards.

The current draft of the AI ​​Act seeks to categorize applications of AI into levels of risk to consumers, with AI-enabled video games and spam filters in the least risky category.

High-risk AI systems include systems used to assess credit scores and access to loans and housing. This will be the focus of tight control over technology.

But as AI continues to develop rapidly, Vestager said we need to be realistic when it comes to fine-tuning the rules around the technology.

“It’s better to get say 80% than not to get 100%, so start, learn, come back and fix it with others,” she said.

Vestager said there was a “definite risk” that AI could be used to influence the next election.

The challenge for law enforcement and intelligence agencies, he said, is to “get the full edge” in the criminal arena, where they risk gaining an edge in the technology race.

“If they can scan your social feeds and get your detailed profile, the risk of being manipulated is huge,” she said. “If we find ourselves in a situation where we cannot believe anything, we have completely undermined society.”

But Vestager said this is not realistic.

“No one can force it. No one can make sure everyone agrees,” she said, noting that some use the suspension as an opportunity to get ahead of their competitors. pointed out that it is possible.

“I think it’s important that every developer knows they’re all signed up for the same guardrails so no one takes undue risk.”

The European Parliament’s AI law proposal aims to limit the use of biometric systems and the indiscriminate collection of user data from social media and CCTV footage for purposes such as facial recognition systems.

But Vestager said, “We put in place tight guardrails so that they are not used in real time, but in certain situations, such as when you’re looking for a missing child or when a terrorist is on the run. We want it to be used only by

“Congress has taken the more principled position of voting tomorrow to basically outright ban it.”

Before the AI ​​law is finalized as the world’s first rulebook for the use and development of AI systems, all three power branches of the EU – the European Commission, Parliament and the Council – need to agree on the final version.

It is not expected to enter into force by 2025.