Artificial intelligence may solve many problems, but without research we will only transfer our bad habits to it

In order to take full advantage of artificial intelligence while minimising the damage, long-term research and public education are needed, says Researcher Teemu Roos.

In 2016 two unexpected and politically significant events took place. In June, British people voted for Brexit, that is to say, to leave the EU. Four months later, businessman and TV celebrity Donald Trump was elected as the president of the United States.

The veracity of these results was doubted, but instead of traditional election fraud, something entirely new was found in the background: Prior to the elections artificial intelligence had been harnessed to spread political propaganda in social media. Some voters had been manipulated with algorithms and bots spreading misinformation.

“These are both examples of the kinds of large-scale problems artificial intelligence may cause.  The best remedy I have come up with for these and other issues related to artificial intelligence is for people to become aware that these problems are possible today,” says Teemu Roos, associate professor at the Department of Computer Science at the University of Helsinki.

Artificial intelligence helps to understand particle accelerators – and fairy tales

Roos is renowned as a researcher of artificial intelligence and is the coordinator of the Elements of AI online course, which has attracted worldwide interest and praise. One of the aims of the course organised in collaboration with software provider Reaktor was to educate 1% of the Finnish population to understand the basics of artificial intelligence.

Studying the political effects of artificial intelligence is not Roos’s specialty, but only one of the many phenomena that interest him. As an artificial intelligence researcher, Roos’s research is linked to many disciplines, from physics to folklore and politics.

“My research is largely multidisciplinary and may show up almost anywhere. I aim to be present in clearing the path between my or someone else’s research and practical applications of artificial intelligence.”

As an example, Roos mentions a study conducted in collaboration with the Helsinki Institute of Physics (HIP), where artificial intelligence is used to evaluate material properties in extreme conditions.

“We can study at the level of atoms how different materials react, for example, inside a fusion reactor or particle accelerator. Artificial intelligence allows us to do so without having to build equipment costing millions or billions of euros.”

In another, very different project, Roos collaborated with anthropologists and historians to study the evolution of fairy tales: how extensively known fairy tales, such as Little Red Riding Hood, have spread in different cultures and what cultural differences apparent in the various versions of the fairy tales tell us about different peoples. In this study, the role of artificial intelligence was to help find connections and similarities between the fairy tales.

Bad education will lead to us transferring our bad habits to artificial intelligence

Solutions to many problems and needs faced by humankind, from translation to driving cars, are currently sought from artificial intelligence. Often the wish is that artificial intelligence will become a source of agency free of the errors and emotions that plague human beings.

However, artificial intelligence will be only as intelligent as the material used to teach it. If artificial intelligence is, for example, taught how to recognise criminals with biased material, it will easily duplicate human prejudices based on gender or skin colour. Artificial intelligence can also be used malevolently, as happened prior to the US presidential election and Brexit.

“If artificial intelligence is a megatrend, ethical artificial intelligence is a megatrend within it. It is important that individuals from as diverse backgrounds as possible participate in researching and teaching artificial intelligence, so that we can minimise these kinds of issues,” says Roos.

According to Roos, teaching artificial intelligence to avoid skewed ways of thinking may be challenging – but not as challenging as teaching humans.

“This kind of thinking can be recognised and then acted upon. When it comes to people, this is much more difficult.”

Finland is a leader in artificial intelligence research

Artificial intelligence research is doing well in Finland. Measured by the number of scholarly publications on artificial intelligence, the universities in the Helsinki Metropolitan Area compete evenly with the top universities of the world. On the corporate side approximately 4,000 companies are linked with artificial intelligence in one way or another, 50 of which have large teams working on artificial intelligence applications. This is a significant number when compared internationally.

Much artificial intelligence research is multidisciplinary. It is often slow and requires long-term funding.

“It may take years for people from various fields to learn to talk with each other. Projects often last a couple of years, and launching them is risky, if there is no certainty about continuing funding.”

Ultimately investment in artificial intelligence is also an investment in much more.

Roos says that it has been proven that universities receive too little recognition for the artificial intelligence research they conduct.

“There is a great deal of talk about research conducted by Google, IBM and Facebook. Yet the individuals working in these companies have been educated by universities.

“Secondly, without the university ecosystem, the majority of companies could not utilise artificial intelligence. Large companies may conduct their own research and product development, but even they don’t have the desire or resources to conduct the critical basic research on which innovations are based,” says Roos.

Artificial intelligence is the word of the day, and in this current boom, decision-makers in politics and the corporate world easily lose sight of the difference between experts and “experts”. Genuine expertise is needed to ensure that the Finnish population can be trained to recognise the possibilities and dangers inherent in artificial intelligence – for example, the potential to shape opinions.

“When it comes to issues related to artificial intelligence, researchers are the experts you should listen to,” says Roos.

Text: Sebastian Koskinen

 

Teemu Roos

Teemu Roos is an associate professor at the Department of Computer Science at the University of Helsinki. Roos’s research focuses on machine learning, in other words, how machines can be taught to reason and independently come to a desired conclusion.

Current topics of interest in the field:

“Artificial intelligence helps to personalise the content, for example, of music or films. On the other hand, artificial intelligence creates bubbles by offering increasing amounts of content that interests us. This may be dangerous, if we no longer receive information that challenges us or opens up new perspectives. Can we find a balance between these two? I’m eagerly looking forward to the solution to this difficult problem."