Governments and companies are pouring money into artificial intelligence, but it’s still an immature technology, say specialists at Science|Business conference
Beware all the hype about artificial intelligence. In fact, experts say, the technology is still in its formative stages – and so Europe still has an opportunity to overtake an early American and Chinese lead in the field.
That’s the view of Petri Myllymäki, professor of computer science at the University of Helsinki. Though news reports of artificial intelligence can sound space-age, in fact the underlying technology of “the current AI is very old,” he says. “It is about 30 years old and it is really stupid,” he told a workshop on AI at a conference on innovation in the EU organised by Science|Business.
“The next generation will be the big revolution. We haven’t seen anything yet,” Myllymäki added, explaining that the next wave of AI systems will help knowledge workers retrieve information, enabling them to focus on more creative tasks.
Governments and ICT companies around the world are pouring money now into developing that next wave of AI technology. In the same week as the conference, the European Commission announced that it will spend €2.5 billion to help spread AI across the European economy and society; EU member states are also investing, with France announcing a €1.5 billion plan. But private sector investments dwarf such numbers. In the private sector, the Commission says, European investments in AI were between €2.4 billion and €3.2 billion in 2016, compared to almost €10 billion in Asia and €18 billion in the US.
But the technology being used is still in its infancy. The commercial AI systems developed by the leading US Internet platforms generally use deep learning techniques to digest large volumes of data and identify particular images, spoken words and other patterns of data. Such systems are sometimes known as black boxes because it isn’t immediately apparent how they reached a decision. “They work in some narrow environments remarkably well, if you have a lot of data,” Myllymäki noted. “They have been productised and there are nice tools you can use, so these are the primary reasons for the current AI revolution.” The next wave of AI will be more transparent, will be able to explain itself and better assess uncertainties.
Can AI cars speed?
The ongoing advances in AI are raising some thorny ethical and legal issues for both software developers and the organisations employing these systems. Antonio Rotolo, vice-rector for research at the University of Bologna, asked the workshop whether AI should be allowed to break the rules and even violate the law, if that is the safest option. He described a scenario in which a self-driving car is on a highway with a speed limit of 40km/h, but all the surrounding cars are driving at 70 km/h. In this situation, he contended that the safest option would be to keep up with the surrounding traffic, “assuming this speed is reasonable given the condition of the road.”
Another challenge is the shortage of people working in AI from diverse backgrounds, noted Lucilla Sioli, director of digital industry at DG Communications Networks, Content and Technology of the European Commission. She stressed the need for more AI developers to have knowledge of other fields, such as psychology and philosophy, while also calling for Europe to involve more women in the development of AI.
Several workshop participants said Europe needs to raise its game in data collection and analytics; curated real-world data is the raw material that AI systems use to learn. “It is extremely important for people in the workplace to treat data in a smart way – they need to be AI conscious,” Raphäel Fabian, EU affairs officer of Rolls-Royce, noted. “That means a lot of up-skilling and retraining” to deal with the data deluge in the workplace. He described how each aircraft flight generates three terabytes of data (the equivalent of about one million digital images) about the performance of Rolls Royce engines.
With the vast amount of digital data now being generated in workplaces every day, traditional user interfaces, such as a keyboard and mouse, could become obsolete. “We need to start speaking to our computers as soon as we possibly can,” stressed Pekka Sivonen, director of digitalisation strategy and programmes at Business Finland, the Finnish innovation agency. “The vast amount of data will just be so overwhelming, there is just no ways and means to put that data in a structured manner behind a UI in any type of machine.”
End data poverty
However, right now, Europe is relatively data poor. Sivonen estimated that “Europe has a market share of 4 per cent of the data that is out there, globally speaking. That is how far behind we are.” He argued that Europe urgently needs to embrace AI to remain competitive with China and the US.
There are some reasons for optimism. Sivonen related how a car manufacturing plant in Finland had increased its workforce from 800 to 4,500 after four rounds of robotisation with AI, which had made the factory more globally competitive. “Is this the kind of development we want to resist? I don’t think so,” he asked, arguing that the combination of humans, AI and robots will create co-bots that can take Europe to new heights of productivity.
Even so, many people fret about the potential for AI to displace human employees, further exacerbating inequality across society. To help allay such fears, Sivonen said Europe should keep citizens at the centre of its AI strategy, and talk about augmented intelligence, rather than artificial intelligence.
“We need to educate people from an emotional point of view to think about AI as we think about our washing machines and our dishwashers,” he added. “That’s the way you are going to feel about AI in five years. It is going to do the most obvious things that nobody wants to do.”