The regulation of artificial intelligence is coming and will encompass liability and damages, European Commissioner for Justice, Consumers and Gender Equality, Vera Jourová said during a keynote at the University College London European Institute June 3.
However, legislation might not materialize for a few years as Europe tries to balance AI investment with regulation, Jourová added.
On the subject of liability and damages, she referred to the Uber Technologies Inc. "autonomous car that killed someone in Arizona" in March 2018. The incident was reportedly the first pedestrian death caused by a driverless vehicle. Uber agreed to a settlement with the victim's family and was cleared of criminal liability over the incident in March 2019. It did not reveal the terms of the settlement.
In tort law, a victim or victim's family can pursue a lawsuit against a defendant if they have been caused harm. To obtain financial compensation, the plaintiff must prove the defendant owed and subsequently breached their duty of care.
"What applies offline, should apply to AI," Jourová said of the EU's approach to legislation.
In the meantime, Europe believes that the GDPR — the stricter data-privacy laws introduced last May — is the "basic fix" that will also govern how AI companies handle data.
"The GDPR rules have to be respected and complied with [by AI firms]" Jourová said. "But you will not find any other regulation [for now]."
The commissioner admitted that Europe is lagging behind the U.S. and China in AI research and innovation.
In an effort to play catch-up, Europe in December 2018 announced €20 billion of private and public investments into AI by the end of 2020. This figure would likely be "considerably higher," Jourová said, as the EU is combining its own funds with money from member states and the private sector.
Jourová said the bloc's AI funding strategy includes "several centers of AI excellence plus a number of hubs in each member state."
The EU in April published its guidelines on developing ethical AI, calling for more human oversight over machine learning systems, secure and private data storage, transparency and unbiased services.
Jourová called these rules, which are not legally binding, Europe's AI "principles." "The way the algorithms are designed must be traceable, must be accountable, protections against bias must be embedded in [them] from the beginning," the regulator said.
In terms of EU's overarching approach to AI, "[currently] the plan is not to overregulate, [to] support AI by smart financing and keep the big brains in Europe and attract the big brains from outside of Europe," she said.
Jourová said she had discussed with Facebook Inc. and Google LLC about the speed of AI development and "what comes first, tech, money or the law?"
"Legislation is not always the best way to solve problems," Jourová said, but she reiterated it was on the horizon.
"We want to see how [the guidelines] work and then in several years time we will address the legal part."