Sci-fi imagery linked to artificial intelligence tells the wrong story
Humanoids and other illustrations inspired by science fiction in connection with stories about artificial intelligence may seem harmless, but using them year after year to illustrate the topic can be detrimental.
Lue tämä suomeksi
“Images associated with artificial intelligence have rubbed me the wrong way for a long time. When I first gave the Introduction to AI course more than a decade ago, I noticed that many students’ notions of artificial intelligence were based on stories by Isaac Asimov and other sci-fi authors. I think someone was even disappointed when, instead of focusing on what robots are allowed and not allowed to do, the course was largely about mathematics,” says Professor of Computer Science Teemu Roos from the University of Helsinki. Roos also heads AI-related education at FCAI.
Images of intelligent robots that potentially threaten humanity, which have their roots in the science fiction of the 1960s, persist in discussions about artificial intelligence. The scope of the matter can be seen by using ‘artificial intelligence’ as a search term on Google. The search produces a slew of very similar blue images: the brain connected to various blue circuits, robots and humanoids in environments associated with space. Image banks also offer similar blue-tinged images.
A collection of typical image bank images that are used to illustrate articles on artificial intelligence, even though their message does not match reality. Image: Mia Paju
“The Terminator continues to keep me busy. Every time I see a terminator, I know I have to show a countless number of other kind of images to make people again forget about terminators,” Roos laughs, continuing his efforts to steer people away from using images that lead down the wrong paths.
“Then again, it's understandable. After all, none of us adults today studied artificial intelligence in school, and it’s still not widely taught even now,” Roos notes.
Clichéd images cause harm
According to Roos, clichéd images of artificial intelligence steer people’s thinking off track, which can cause a lot of harm.
“Clichéd images prevent people from seeing how artificial intelligence is already being used almost everywhere. AI solutions are reshaping our environment and affecting our public administration, politics and free time as well as how we encounter or do not encounter one another,” Roos says.
Images of humanoids can alienate potential experts in artificial intelligence and people whose voice should be heard in the discussion on artificial intelligence.
“If we see nothing but brains with electric circuits, such a catalogue of images excludes everyone except those who are into circuits. Sci-fi imagery evokes a lot of antipathy. For instance, my wife immediately changes the channel if she sees sci-fi on the television. She’s a doctor, and you would hope that doctors in particular would be interested in artificial intelligence,” says Roos.
Roos also highlights legislation and points out that laws pertaining to artificial intelligence must be up to date.
“Politicians are humans too, and they have the same misconceptions as other people. A politician thinking that AI is a futuristic phenomenon and wondering what would be required of them if such a phenomenon were to be developed at some point is a terrible way of embarking on drawing up legislation, because artificial intelligence has been a significant force of change in our society for at least 10 years.”
International collaboration to change AI imagery
Roos is not alone in his thoughts. Better Images of AI, a collaboration project established last year, is calling, on a wide front, for better illustrations in stories about artificial intelligence; images that also include the human side of technology and that would describe the functioning of artificial intelligence more realistically.
A number of influential parties, including BBC Research & Development, the Alan Turing Institute and other research institutes, NGOs and universities are contributing to the Better Images of AI project. The Finnish Center for Artificial Intelligence has also joined the project.
“This project is a perfect match for FCAI’s goals. We want better AI imagery in order to attract many different groups of people interested in AI, understand it more broadly and take part in related discussion,” Roos says.
The goal of the non-profit collaboration is to investigate, create, curate and offer better images of artificial intelligence. The project website lists proposals for better images for illustration purposes. At the same time, the aim is to encourage image providers around the world to start producing images themselves that better correspond to the reality in which we live.
Of course, illustrating topics related to artificial intelligence is not always easy, but Roos offers as a ground rule the idea that the images should also include people. He likens illustrating artificial intelligence to illustrating news about the economy.
“Like the economy, artificial intelligence is an abstract topic, but not all releases concerning the economy are illustrated with money and graphs. Instead, the illustration is chosen on the basis of the content of the news item, the people it touches, and how the matter is evident in people’s everyday lives. The same rule can be used when illustrating stories about artificial intelligence.”
What about journalists – have they already found better AI images?
“Yes they have, at least those who interview me. Every time I give an interview on artificial intelligence to a media outlet, I warn them not to use terminator images if they want to interview me again later,” Roos chuckles.
Teaching the public about artificial intelligence
Teemu Roos has a lot on his plate with regard to education related to artificial intelligence. Roos is the coordinating teacher of the popular Elements of AI online course. The course is a central part of Finland’s AI strategy, according to which artificial intelligence belongs to everyone. The online course has already attracted more than 820,000 students from around the world.
Roos also heads AI-related education at FCAI. In addition to academia, Roos is interested in school education in Finland. He is in charge of the Tekoäly ja lapset (‘AI and Children’) project, which introduces teaching related to artificial intelligence to primary school.
But let’s get back to those AI images. What does artificial intelligence actually look like?
“People. Either those involved in its development, or those whose lives it affects. Another important element in this is the appearance of these people. They don’t all look like me, for example.”
Text by: Mia Paju
Justification for poor imagery
Notions of the beauty of images vary, and creating aesthetically pleasing images is not always possible in a hurry, but why are the following illustrative elements a poor fit for stories about artificial intelligence? Teemu Roos has the answers.
Brain – Intelligence has been thought to reside in the brain, but the idea that artificial intelligence would somehow come to resemble human intelligence is misleading. Rather, artificial intelligence is a tool, a hammer or a pocketknife, that can be used to do many things.
Space – Space distances us from our environment and people. In most cases, the use of artificial intelligence is about making people’s lives easier or improving our environment, not about distant space.
Robots – Robots can be used in stories that are about the specific robot in the illustration.
The colour blue – Traditionally, blue has been associated with machines and electricity. Blue is also a cold colour. This combination separates the human aspect of artificial intelligence, especially if such blue images contain imaginative futuristic elements.
Electric circuits, lights – This is traditional nerd imagery that attracts electrical engineers and readers of technology magazines. Personally, in spite of working with technology, I perceive artificial intelligence from another perspective. I am fascinated by algorithms and mathematics.