Site icon European Science-Media Hub

Demystifying artificial intelligence

Natural Language Processing - NLP - Speech Recognition - Computational Linguistics Concept

During the STOA workshop ‘AI Public Perspectives’ on 14 November in the European Parliament in Brussels, Geertrui Mieke De Ketelaere will give a keynote lecture about ‘Demystifying artificial intelligence’. She talks to the European Science-Media Hub about her view on the public perception of artificial intelligence.

‘AI is about algorithms’, said a computer scientist. ‘AI is about machines that can learn from data’, said an engineer. ‘We need to do something with AI’, said a business manager. ‘AI is an existential threat for humanity’, said a sociology student.

Over the past decade, artificial intelligence has become branded as AI, but does everyone mean the same when they speak about AI? Not really. “A Babylonian confusion of tongues has arisen”, says Geertrui Mieke De Ketelaere, Adjunct Professor at Vlerick Business School in Ghent, Belgium.

Geertrui Mieke De Ketelaere: I started working in the field of artificial intelligence when it was just an academic discipline, in the early 1990’s. A decade or so later, businesses started to be interested in AI, and they started to use their own language, stating things like ‘data is the new oil’. Later, some ten years ago or so, governments stepped in and started using their own language around rules, laws and ethics. And now, scientists, technologists, business managers, sociologists and lawyers are sitting around the table, talking about AI, and they don’t understand each other any longer.

So, what should be done?

Geertrui Mieke De Ketelaere: We urgently need ‘AI translators’, people who translate the needs and demands of engineers to managers, or the concerns of citizens to developers. We need people who can talk about all aspects of AI, people who can connect people, planet and profit – the social, environmental and economical domains. We should adopt AI when it has clear added value, but not when it doesn’t.

Why do you also refer to the planet-part when talking about AI?

Geertrui Mieke De Ketelaere: Because AI is going in the wrong direction in terms of energy consumption and its CO2-footprint. At present the CO2-emission of AI- and internet of things-applications combined is about two percent and it is growing.

[For comparison: this 2% is roughly the same as the CO2-footprint of the airline industry.]

How do you look at the public perception of AI?

Geertrui Mieke De Ketelaere: Citizens are very important. At the end they decide whether or not they accept a certain piece of technology. Take the Google glasses. People didn’t accept them. The designers forgot to include the community when they were developing the glasses. And still, when you hear talking about building something ‘by design’ into some application, very often only the techies are at the table.

Having said that, many people just want to get started right away with gadgets, but don’t want to know how they work. They accept cookies, they start using a digital toothbrush without reading what the impact on their privacy is.

What are your ideas about influencing public perception of AI?

Geertrui Mieke De Ketelaere: My way is to explain factually how AI works, as much as possible in plain language. Take the example of an AI-system that recognises the handwritten digit ‘8’. We humans recognise the digit straight away. A computer vision system only sees pixels. And every pixel has a number ranging from 0 to 256, representing grey scales between white and black.

Another example is a computer vision system looking at the picture of a snowman. If I turn the snowman upside down, the system doesn’t recognize it any longer, whereas a child would. In other words, the AI-system has not learned the abstract concept of a snowman.

There are many such examples that show the difference between human and artificial intelligence. In some aspects AI is better, for example in the amount of data it can learn from, but there are many other aspects of intelligence in which humans excel: creativity, consciousness, common sense, emotions, abstraction, generalisation, causal reasoning.

What role can the arts play in the public perception of AI?

Geertrui Mieke De Ketelaere: Explaining how AI works in technical terms can be boring for some people. The arts can polish away that dullness and thus function as an ‘AI translator’.

I collaborate with the Belgium artists Dries Depoorter, who creates apps, games and interactive installations. He created an installation called ‘Data Broker’, which assembles, shares and experiments with private data of himself and random people he found on the internet. The installation raises awareness about issues like social identity, data sharing and online privacy. It visualises what I can not visualise when I write about AI in my book. That is an added value of the arts.

What, according to you, does not work well if the arts try to speak about AI?

Geertrui Mieke De Ketelaere: Science-fiction stories in which robots take over the world are not effective. This is so far away from reality. Some people say that science fiction holds a mirror up to us, but I doubt that. I have never seen a serious debate arising from such far-fetched stories.

A lot of science-fiction stories are based on anthropomorphising AI, whereas, as you said, AI and human intelligence are quite different…

Geertrui Mieke De Ketelaere: When AI was founded as a scientific discipline in 1956, the dream was to build humanlike intelligence in a machine. But while humans were working on realising this dream, humans also started to get scared of what was being created. I think that this is the result of our arrogance of thinking that we are superior to animals, and definitely superior to plants.

I can get tears in my eyes watching a gymnast, but her performance has nothing to do with mathematics or logic. It is a fantastic form of intelligence in her body. I recently learned that my favourite perfume, Chanel No. 5, is based on a type of jasmine smell that plants use to communicate to other plants when there is danger. While we dream of building intelligence in a machine, we don’t want to see that there is so much intelligence in nature we haven’t explored yet.

Useful link:
• Artificial Intelligence (AI): hindsight, insight, foresight
• STOA workshop on 14 November 2022: Artificial intelligence public perspectives

European Research Council Projects:
“AI all around us”
Agnieszka Wykowska is a senior researcher of the Social Cognition in Human-Robot Interaction at the Italian Institute of Technology (Istituto Italiano di Tecnologia) in Genova, Italy. In 2016 she has been awarded an ERC Starting Grant for her project “Intentional Stance for Social Attunement” whose goal is to investigate if humans are ready to engage in social interactions with humanoid robots. Dr Wykowska will present her findings at the ERC’s conference Frontier Research and Artificial Intelligence.

“Socialising with Artificial Agents”
A number of factors have played an important role in the evolutionary success of the human species. One of the undeniably fundamental factors has been our inherent ability to communicate. This capacity to perceive, respond to and coordinate behaviour with others has not only allowed us to survive, but also to thrive. The ERC-funded project SOCIAL ROBOTS headed by Prof. Emily Cross is aiming to gain a deeper understanding of the intricacies of how we comprehend and coordinate our actions with other people and with robots to achieve mutual goals.

Exit mobile version