Interview with MIT researcher Stefania Druga: How children understand artificial intelligence

SPIEGEL ONLINE: Ms Druga, a new generation is growing naturally with AI applications. Do children perceive these as more of a family member or as a tech object?

Stefania Druga: It depends on the age of the children. There are many studies that show that younger children between the ages of three and seven can not tell well from a social robot or an animated object, whether it's a living being or a machine. Smart toys that talk or run are perceived by them somewhere between the person, the animal and the object, that is, neither alive nor dead. As children grow older, from about seven years upwards, they realize that it is only a device - themselves when it shows or speaks about feelings.

SPIEGEL ONLINE: You work and research in many countries. What cultural differences characterize the handling of artificial intelligence?

Druga: Language assistants like Alexa and smart toys are used less often in other countries than in the US - so the perception, understanding and handling of these devices is different. It's something new for kids who do not have such devices at home, and the applications are not available in all languages. Some cultures are also more sensitive to issues such as privacy and privacy.

SPIEGEL ONLINE: Like Germany?

Druga: Especially in Germany, children trust the devices the least. When they came to my workshop, they expected that Alexa would not answer their questions honestly, but try to deceive them. So they already had a negative attitude towards technology through the media and society. The kids did not necessarily interact with the devices differently, but they interpreted the answers differently. For example, they had the feeling that Alexa is lying and can not even know who is President of Germany - because she does not come from Germany.

SPIEGEL ONLINE: Is the skepticism good or bad?

Druga: It's not necessarily a good phenomenon. Media often presents such devices with a fear-driven narrative, but it's more complex. I'm working to democratize this technology: Everyone should be able to talk about artificial intelligence, use it, and understand how it works. This also includes recognizing in which cases it can do something positive and in which cases it is better to do without it.

SPIEGEL ONLINE: Do many parents already set their children rules on how much time they can spend, for example, with AI objects?

Druga: I do not think that's common. There are parents who teach their children to say "thank you" and "plea" when talking to language assistants. But in general, there are less strict rules so far than "Screentime", so how long children are allowed to use the TV or the smartphone.

SPIEGEL ONLINE: What role do parents and the social environment play in general?

Druga: In a study in the United States, we explored how children perceive and interact with Artificial Intelligence. Clearly, children whose families had low and middle income were initially better at working together. However, they had difficulty getting ahead because they had less experience in programming and interacting with these technologies. Children from a wealthier environment initially had difficulties working together but showed a better understanding of AI concepts.

SPIEGEL ONLINE: What needs to be done to reduce the social gap in this area?

Druga: Artificial intelligence needs to be disenchanted. Children are influenced by the attitude of other people in their environment to the technology. But the perception and interaction of children with Artificial Intelligence changes as they learn to program them. Once children and parents understand that they can train the computer or voice assistant in an ongoing conversation, it demystifies how smart these devices are. They understand that the algorithm and the data behind it have been developed by humans. Once the process is disenchanted, parents and children learn how to use the devices to do what they want.

SPIEGEL ONLINE: How can children be taught this very complex technology?

Druga: It has to be fun. If you want to talk to children about autonomous cars, it's boring for them. Artificial intelligence is also much more than Alexa - which is just a commercial product. Although there are many smart toys, even children without access to Alexa or an AI robot like Cozmo need to have the opportunity to acquire such skills. That's why we're also developing open source tools for kids. With the platform "Cognimates" they can develop programs for networked objects, ie the Internet of Things, but also web-based applications.

SPIEGEL ONLINE: How does that work?

Druga: If you want to play with the computer, for example, you have to teach him a game like "scissors, stone, paper" and train him to recognize photos of a stone or a pair of scissors. The more often they play the game, the more often they can give the computer feedback as to whether they are doing something right or wrong. This makes the kids fun and shows how a computer is trained to recognize a photo.

SPIEGEL ONLINE: To what extent mediate such applications also a critical handling of artificial intelligence?

Druga: Children also learn that the computer only recognizes white hands if you only train them with pictures of white hands. They understand such deep issues of distortion and fairness, but in a very tangible way. You realize that if you want to play with your friends, it's important that the application recognizes everyone's hands - so everyone can play together.

SPIEGEL ONLINE: So you do not have to worry about the AI ​​generation?

Druga: Children understand technology much more differentiated than adults and are much more open-minded. They understand that technology is important in the future and that we need to understand how it works - even if there are risks and we need to discuss what we want to automate and what we do not.