Europe debates AI - but AI is already here

"AI has become a major sight of focus of interest, in the media and elsewhere," said Allan Dafoe, director of the governance of AI programme at the University of Oxford, at a hearing in the European Parliament.

"Much of this interest is driven by hype and misunderstanding, but it is my belief that the critical insight that AI will be an important technology of this century is correct. In fact I would argue that how we build advanced AI will be the defining development of the 21st century," he added.

It is clear that Europe is looking for a way to benefit from the advantages that AI can bring, but also that it will try to shape the development of AI to mitigate some of its possible negative effects.

Why now?

So why is AI such a prominently discussed topic all of a sudden?

As several speakers of the recent conferences have said, artificial intelligence itself is not new.

Vladimir Sucha, director-general of the EU's Joint Research Centre (JRC) joked that when he was still at university "almost a hundred years ago", there was already a department of artificial intelligence.

The Slovak scientist graduated in Bratislava in 1985, acquired a PhD in 1991, and taught at the Comenius university in the second half of the 1990s.

But he had a point: the term AI has been around since 1956.

What changed in recent years, is the combination of smarter computers with vast availability of massive amounts of data.

"The past few years have seen rapid progress in various fields that are generally identified under the header of AI," Tim Sweijs, director of research at The Hague Centre for Strategic Studies told MEPs.

"That progress has largely taken off because of advances in deep learning based on neural network pattern recognition," he added.

Super-intelligence far away?

Like other scholars, Sweijs made a distinction between three types of AI: narrow intelligence, general intelligence, and super-intelligence.

A system that has narrow intelligence is very good at a specific task, for example preventing spam messages from entering your email inbox.

"We are currently mostly at artificial narrow intelligence applications," said Sweijs.

"We are far from a situation where there is artificial general intelligence, let alone artificial super-intelligence."

He was supported by Sucha, who spoke at a Visegrad 4 (V4) conference at the Slovak mission in Brussels – the V4 being a loose coalition of the Czech Republic, Hungary, Slovakia, and Poland.

"Is artificial intelligence the future? No, it is not. Artificial intelligence is the present. We already have artificial intelligence on a daily basis," said Sucha.

"Maybe not that intelligent as we see in science fiction movies, but it is with us," he added.

Experts disagree over when we can expect an intelligent system that matches or exceeds human-level intelligence.

"By when might we see radically transformative capabilities, or even what has been called artificial general intelligence – broadly, human-level AI? The short answer is that we just don't know," said Dafoe.

"Anyone who confidently predicts when it will occur, either knows something that they're not telling us, or is not well-calibrated in their probabilities," he said.

Will robots take my job?

A panel discussion on how AI will change employment, at the V4 conference, was virtually unanimous in its agnosticism.

"We don't really know what will happen in the future. It is not a scientific debate, but a debate of faith," said Robert Pinter, sociologist and assistant professor at the Corvinus University of Budapest (Hungary).

"It is very difficult to estimate the impact of artificial intelligence on the labour market. We still don't have enough data and experience," added Dominika Kaczorowska-Spychalska, assistant professor at the university of Lodz (Poland).

More research is needed, said Stijn Broecke, but there is reason for optimism.

"Artificial intelligence will probably create lots of jobs as well. New technologies have always destroyed jobs, but they have always created more, and often better, jobs," said Broecke, senior economist at the Organization for Economic Co-operation and Development (OECD).

He also echoed a remark made by JRC chief Sucha – who had said: ""The future is not written yet. … We can shape it in a way which is European, but we need to be active, we need to be forward-looking."

"As a society we make decisions, ethical decisions about where we apply artificial intelligence and where we do not," said Broecke.

We may choose that even though some tasks can be full automated – like nursing or teaching – they should not.

No time for long debates

One problem is that the challenges and opportunities provided by AI may require a lot of time to digest, but that time is not there.

Sucha told the Brussels audience that even though they all "love … to have all our beautiful bureaucratic processes", there was no time for that.

He was backed up by a participant at a separate debate, held in Brussels under the so-called Chatham House rules – meaning that information cannot be attributed to the speaker.

"While we are debating, US and China are doing it," the participant said.

Nevertheless though, that is still how the EU, a collection of semi-sovereign states, works.

Radim Polcak, head of the Institute of law and technology at the Czech Masaryk University in Brno, sat down with EUobserver for an interview.

Polcak had spent some six months as an advisor to the cabinet of Czech EU commissioner Vera Jourova, in the run-up to the commission's strategy paper on AI published last April.

He praised the civil servants working at the commission's various directorate-generals and the Jourova cabinet.

"They have very high level expert understanding of the issue. That was my very positive surprise," said Polcak.

"The problem is very often that when you do this on a European level you have to have compromises. Very often people who have a high level of skills, they can't use their skills to the full extent because it is limited by the need to reach compromise in different areas."

The EU is trying to achieve these compromises. The commission is currently working with the national governments of the EU member states on a coordinated plan about AI, due to be finished before the end of the year.

The idea is that Europe can offer a middle ground between the US and China, and help shape the values that will be programmed in the AI systems.

This is the first in a three-part series of articles on how the EU deals with artificial intelligence.