Impact on Children

[milessea]

Contribution by the W3C Community Group on Accessibility for Children:
https://www.w3.org/community/accessibility4children/
You are welcome to contact chairs for more information.

Key messages of this contribution:

• Proportionality principle: 1 in 3 web users are children .
• Protect and empower children on the web.
• For children with specific needs, accessibility is needed as of early stages when involving children for design.

3 sections of this document:

1. The context of digital rights and protection for all children
2. Children and accessibility -Key Performance Indicators
3. Topics addressed by W3C with regards to children with specific/ functional needs

1. The context of digital rights and protection for all children
Our Community Group would like to highlight the importance of protecting children in the online world.

The convention on the rights of the child has added a specific comment to address the children’s rights online specifically: General comment on Children’s rights in relation to the digital environment

The UNESCO recommendation Recommendation on the Ethics of Artificial Intelligence [UNESCO-AI] referred to in the W3C paper on Ethical Principles for Web Machine Learning highlight the need to
· protect vulnerable populations and age -based vulnerabilities. Specifically in the fields of mental health, education and robot / chatbot use by children. With specific attention to anthropomorphization ( to attribute human form or personality to things not human), when children are involved.
· “promote collaborative research into the effects of long-term interaction of people with AI systems, paying particular attention to the psychological and cognitive impact that these systems can have on children and young people”.
· “put in place mechanisms to meaningfully engage children and young people in conversations, debates and decision-making with regard to the impact of AI systems on their lives and futures”
These topics have been addressed in our sessions on accessibility for Children as very important components of our work on accessibility with children.

Cifar (a Canadian-based global research organization) issued a policy brief including an overview of the literature on children’s rights for responsible AI: https://cifar.ca/wp-content/uploads/2024/04/CIFAR-Responsible-AI-and-Children-EN_Final.pdf

Proportionally, children are important actors on the web: “Globally, 1 in 3 internet users are under the age of 18 years.”.
The report formulates recommendations in order to build a responsible AI ecosystem for children acknowledging that children are a very diverse population.
Children of all ages should be “involved in research, policy decisions , technology development and designs that are going to impact them”.
The report also provides a summary of research in four areas:

1. everyday life, play and learning.
2. Social relationships
3. development and well-being
4. Privacy and other rights

The report also mentions that “children should not simply be protected but also empowered in their interactions with AI.
Less attention has been paid, however, to practical strategies for involving children in technology development.”
Hands-on activities are suggested at the end of the report.
Our group on accessibility for children found the report very interesting as a state of the art analysis and recommendations. The activities at the end, however, would need some accessibility adaptations before being able to really empower children with specific needs to participate to their full potential.

2. Children and accessibility -Key Performance Indicators
Children with disabilities are very often ‘invisible users’ on the web.
While AI can help foster personalised experiences, there are still privacy issues at stake.
For children with specific needs our Community Group on Accessiblity for children has identified KPI’s in order to evaluate if web content / interfaces have a positive impact on their inclusivity, and accessibility.
These indicators are also useful to measure the performance of AI systems in terms of impact on children with disabilities. Our Community group is reviewing these periodically.

1. Social considerations (peers, relations)
2. Balance Safety - Agency (in all environments emergency contexts, child alone or in care lack of consistent parental authority for consent)
3. Impact on children with specific needs (statistical relevance, scalability …)
4. Awareness assessment (what does the child need to know, how, why)
5. Assistive technology: knowledge present, learning readiness or staged capacity building
6. Complex/ intersectional / evolving needs
7. Transposable (in different contexts)

3. Topics addressed by W3C with regards to children with specific/ functional needs
The topics of the W3C report are addressed for children with disabilities specifically.
Regarding transparency, children with disabilities also need to be aware of computer AI-generated content. The accessibility means might differ from adults as their literacy is still evolving as are their capacities to learn assistive technology.
Privacy and user control are also topics under development for children and children with specific needs would benefit from fine-grained control over their data if this were provided in an accessible way with age appropriate consent management and accessible formats.
A risk -based approach regarding multimodal approaches, models and systems can have a positive impact on accessibility, giving the user many options to choose from. This huge opportunity in terms of accessibility needs to be balanced with safety, security and privacy issues. For children with specific needs, who are considered more vulnerable and at risk of cognitive manipulation, the mitigation should be very robust and evidence based with careful consideration of psychological and social (community impacts). Training data coming from children especially when identifying specific needs should be addressed with caution as sensitive personal information or special category data.
In terms of equity, Children with specific needs should get the same opportunities. Many AI based tools on the internet are focused on ‘problem detection’ or ‘diagnosis’ and do not always develop further in order to support accessibility tools and or personalized learning tools. Tools need to be developed in response to the needs. Further research is needed to show if there is enough support in STEM areas which seem less developed than Literacy (dyslexia detection and support).
See European Journal of Science, Innovation and Technology 2024 https://ejsit-journal.com/index.php/ejsit/article/view/397/373