Safe AI for children is urgently needed

Signal post, January 2022.

There is a tendency to consider mainly adults in various AI related publications and official documents. However, AI has become part of children’s daily lives as well, and this interaction is affecting them in ways we may not fully comprehend yet.

Recently, Amazon’s intelligent virtual assistant Alexa instructed 10-year-old girl to touch live plug with a coin. Luckily, the girl did not get hurt because the mother was present and intervened immediately. The ”challenge” was presented by Alexa as simple and fun. In other circumstances, if left alone with such ”assistants”, children could suffer serious consequences.

AI must not, under any circumstances, jeopardize the most vulnerable – our children. Such risks should never be allowed in AI powered devices: they may look cool but could easily make use of children’s natural curiosity and trust, while directing them towards potentially life-threatening danger. This issue is complex. On the one hand, children need to be aware not to blindly follow such systems’ guidance for entertainment. On the other hand, the so called “challenge” selected by Alexa “from the web” has been created by humans, and allowed by the tech companies to circulate. There are no layers of protection; there are no proper regulation and oversight. Safe AI is urgently needed for children.

UNICEF’s Office of Global Insight and Policy is leading a two-year project to better understand how AI systems can protect and empower children. A month ago, UNICEF held a Global Forum on AI for Children to discuss practical approaches to child-centred AI policies and systems with experts from around the world. This initiative identified an important gap:

“Most national AI strategies and major ethical guidelines make only cursory mention of children and their specific needs. For country policies, references to children are most often in the context of preparing them to work in an AI-centric economy. But as children increasingly use or are affected by AI systems in everyday situations, the lack of attention on the opportunities and risks that AI systems hold for children is growing.”

The global calls for implementing ethical principles and human-centred approach to AI shall not only be applied to AI designers and established tech companies: they shall also be tailored to society’s little citizens, the children. According to UNICEF, it is timely and important to shift from general human-centred approach to include child-centred lens to enable protection, provision and participation by all children on matters concerning AI and ultimately, their future. We need to educate children about AI risks and harms, as well as about possibilities to use AI to improve quality of life. UNICEF’s Policy guidance on AI for children and the concrete tools to operationalize the UNICEF policy guidance on AI for children are important steps in this direction.

Alexa, are you listening?

The writer: Nadezhda Gotcheva (VTT)

Picture: Pixabay