Currently reading: Aalto Leaders' Insight: 5 + 1 Myths about Artificial Intelligence
Previous page Otso Alasko
Digitalization

5 + 1 Myths about Artificial Intelligence

Few subjects in business and technology are causing as much interest and excitement – or fear and distrust – right now as artificial intelligence. Aalto Leaders' Insight talked with Teemu Roos, Associate Professor at the Department of Computer Science, University of Helsinki, about a few common myths about AI.

Annamari Typpö, 18.10.2019

|

Articles

1. The biggest benefits of artificial intelligence in business come from cutting costs

"If you look at the situation over a one-year horizon that may be true, as it's too short a time span to do much else than to improve the efficiency of your current operations," Roos says.

"Cutting costs may be necessary and give you a significant competitive advantage, but if you take a longer-term view, the situation will be reversed. Those who come up with whole new business models will wipe the floor with those who just keep doing the same old thing."

AI can be described as an innovation catalyst."

According to Roos, the true benefit of AI is that it will make the impossible possible. With AI, you can develop services and business models that you wouldn't even have thought of before. AI can be described as an innovation catalyst.

Amazon, for example, would never have been the success it is today, had it not been for AI. It was one of the first to use AI technology to drive its product recommendation engine, which generates 35% of the company’s revenue. "That gives it a unique competitive advantage," Roos says.

Another example he gives is Uber, which has become a huge success, thanks to artificial intelligence. Uber and other sharing economy giants use AI in their core processes, such as seller/buyer matching, pricing, and fraud prevention. Without AI, these companies couldn't operate efficiently enough nor provide the desired level of service to their users. AI is the key to their competitiveness.


2. Humans, not technology, are the biggest obstacle to the large-scale utilization of artificial intelligence

"Yes, that's exactly true. We already have an enormous amount of untapped potential when it comes to technology. There is, of course, a constant need for new research and pushing the boundaries, but the truth is that applications aren't nearly at the level where they could be," Roos says.

According to him, this is due to lack of knowledge and talent. There's a huge gap between what is possible and what we can do.

Another human-related limitation has to do with regulatory maturity. "The rules regarding data protection, data collection, and online privacy must be clear to everyone, so that there is no backlash causing people to refuse to use modern applications."

Roos says we should focus on learning all there is to know about AI now, not wait for it to improve. "There's no point in biding our time until it one day reaches its peak. That's not going to happen."

3. When it comes to the ethical issues of artificial intelligence, we focus too much on its military use

"I don't think we do. In my experience, the discussion revolves mostly around privacy issues, and that's a good thing," says Roos.

According to Roos, privacy issues are important, for example, in light of recent incidents related to misinformation and voter manipulation on social media, made possible by AI algorithms, automation, and big data.

Screen addiction is a particularly pressing issue because it applies to our children, too."

Another ethical problem Roos mentions has to do with how mobile app developers use AI-based optimization techniques to create and feed our mobile addiction.

"Spending too much time doing things that aren't good for us is a genuine problem that I'm sure many of us can identify with. That's what we should be afraid of, not some AI killer robots about to do us harm. Screen addiction is a particularly pressing issue because it applies to our children, too," Roos emphasizes.

Roos is concerned about commercial and political manipulation made possible by the utilization of big data. "Our personal information is used to influence us and drive polarization. In my opinion that's one of the biggest threats related to AI."

Roos doesn't believe that industry self-regulation is enough to combat the threat. "It's quite obvious when you look at what's going on. We need governmental or intergovernmental regulation to coerce everyone to follow the rules. EU's GDPR regulation is a good example of this."

4. You must be mathematically gifted to understand artificial intelligence

This myth is based on a misconception, says Roos.

"AI and mathematics in general have long been plagued by the myth of the lonely genius when the truth is that research and innovation always happen in teams. If you're part of a team building an AI solution, you don't have to be the one who understands each and every math concept. Often, it's enough to know the basics so that you can communicate with the programmers. Marketing specialists or psychologists, for example, may be just as important innovators in an AI project."

Roos also points out that if you focus too much on innate talent you forget that behind so-called talent lies an enormous amount of work and a willingness to learn.

If you're part of a team building an AI solution, you don't have to be the one who understands each and every math concept."

"I would argue that while not everyone can win the Nobel Prize in particle physics because it takes time to acquire all that knowledge, anyone can learn coding or math or AI – and not just when they're young but at any age. The Elements of AI online course is a good example. A number of people over 80 have taken the course successfully."

The importance of knowing enough about artificial intelligence is not limited only to those who build AI systems and solutions, Roos says.

"We all live in a world where we use AI or are being influenced by it. That's why it's really important that everyone can form an informed opinion – not based on hearsay – so that we can engage in civic dialogue to create a common code of ethics of what's acceptable and what's not. In a democratic society, we can't leave it to the hands of elite coders."

5. Europe can't keep up with China and the United States in the AI race

"This is a purposefully built myth designed to attract talent and investments to China and the US," Roos says.

He admits that there's a bit of truth in it, as there always is in legends, mainly because of large investments in innovation by Europe's competition.

"In Europe, we're often quite rigid and don't move as quickly. On the other hand, we have a lot of intellectual capital and the quality of our research is high. I believe that we compete on the same level, if not higher, when it comes to the quality and quantity of AI research."

Roos says that China and the US are usually compared to individual European countries, giving them an unfair advantage.

I believe that we compete on the same level, if not higher, when it comes to the quality and quantity of AI research."

"When you treat Europe as one entity, the situation is different. Where we have room for improvement, though, is promoting and marketing our expertise."

Roos also points out that Europe has enormous potential, thanks, for example, to our high level of education.

"Our situation is far from dire. We just need to focus more on innovation and turn our expertise into business. This calls for national AI strategies and AI hubs to ensure that we have the critical mass to compete with China and the US."

When it comes to regulating artificial intelligence, though, Europe is in the driver's seat, Roos says.

"We have chosen to focus on so-called human-centered AI, which is based on European values. It's about increasing wellbeing and respecting human rights while also providing support and guidance for business. It might not be the most competitive approach in the short term, but it's what's right."

+1. Artificial intelligence creates more jobs than it destroys

"That seems to be true," Roos says. "When studied more closely, estimates regarding the number of jobs that AI will destroy have turned out to be grave exaggerations".

According to Roos, automation will replace specific tasks within jobs, rather than entire occupations themselves. Radiologists are a case in point. There's been a fear that robots will soon take over from radiologists because they are more efficient in reviewing and interpreting medical images, but that's only a small part of a radiologist's job.

"It feels as though the estimates have been done without knowing what people in certain jobs actually do. There's a huge potential for improving efficiency in nearly every occupation, of course, but rather than serving as a replacement for human brain, artificial intelligence is more a supporting tool that we can use for our own benefit." 

Teemu Roos is one of the instructors in the Diploma in AI program, a joint effort between Aalto PRO, the University of Helsinki Centre for Continuing Education HY+, and the Finnish Center for Artificial Intelligence FCAI. Read more about the program


Back to Aalto Leaders' Insight main page

Find more content on