Search
  • News

AI & Society workshop explores the risks of AI on children

by Krista Davidson Apr 30 / 19

A CIFAR-led AI & Society workshop will explore how AI could shape the development of children, and what can be done to minimize the potential risks and create new opportunities for children growing up in the digital age.

Young people account for one in three online users, and AI and digital technology could have significant effects on their development. However, there is little research dedicated to understanding what these effects may be.

_RM14457-3378713569-O (2)“My team has tracked adolescents on their mobile devices over the last decade and witnessed the rapid uptake in device use, screen time and online engagement. At the same time, we have seen increasing concerns about what this constant connectivity means for children’s health,” says Candice Odgers, co-director of CIFAR’s Child & Brain Development program and a developmental psychologist at the University of California, Irvine.

She will lead Generation AI: Reducing inequality and enhancing digital inclusion via smart design and developmental science. The workshop will include an international team of researchers with expertise in machine learning, pediatrics, psychology and more.

Other team members include Anna Goldenberg (Vector Institute, University of Toronto, SickKids), Ronald Dahl (University of California, Berkley) and Mizuko Ito (University of California, Irvine).

One area of interest for the team is the use of algorithms in children’s online content. Children growing up in different family contexts and neighborhoods may be exposed to different and less suitable content that could affect their educational or social development. Particularly when adults are less involved in monitoring and supporting their online activities.

“How do algorithms weigh and use information about a child and their family’s previous viewing, purchasing patterns and postal code?” says Odgers. “Personalization algorithms are designed to maximize time spent online to facilitate advertising and marketing; they are not designed with children’s developmental or learning needs in mind.”

She hopes the workshop will develop new partnerships between researchers, technology companies and other key stakeholders to change this. “Can we build better algorithms that provide equal learning opportunities, protect privacy and optimize opportunities for all young people regardless of family income or post code?”

The team will also focus on building frameworks to support children’s and algorithmic rights in the digital age – ensuring equitable access to online resources, skill building and opportunities, while protecting rights for children who are too young to consent or to understand the meaning of consent.

Personalization algorithms are designed to maximize time spent online to facilitate advertising and marketing; they are not designed with children’s developmental or learning needs in mind.

“The Generation AI workshop recognizes that the internet was not created for children, yet five out of seven children are using the internet,” says Anna Goldenberg, a CIFAR fellow in the Child & Brain Development program, faculty member at the Vector Institute and Varma Family Chair of Medical Bioinformatics and Artificial Intelligence at SickKids.

“In some ways children are a vulnerable population because they aren’t able to consent to much on the Internet. They are the recipients of other people's decisions,” she says.

“We want to consider how it will affect them as they’re developing, and whether it creates disparities or enlarges the disparities that exist already. Is there anything we can do as a society to minimize this impact?”

The workshops will also look at the use of child data in the healthcare system. Robust data sets are needed to tailor health treatments for children, yet information for children is often missing. While this is also true for adult health information, the problem is compounded in children. The workshop will explore solutions for sharing data across multiple jurisdictions and building better algorithms for supporting child health while respecting privacy.

Generation AI is one of four workshops under the first call for proposals for the CIFAR AI & Society program, which aims to examine the questions AI will pose for society, including the economy, policymaking, philosophy and the law.

The workshops are designed to give interdisciplinary research teams project-based funding to explore fundamental topics important to AI and society. The first Generation AI workshop will take place at the World Economic Forum, May 6-7, and will include key stakeholders such as UNICEF, LEGO, YouTube Kids and MIT Media Lab to identify and map out high priority areas for research.


Other resources:
Algorithmic Rights and Protections for Children
Optimize Algorithms to Support Kids Online, Not Exploit Them