Fwd: Web for Children Accessibility & WAI-Adapt

Forwarding to the Adapt list for archiving and sharing purposes.

---------- Forwarded message ---------
From: Maud Stiernet
Date: Tue, Jun 20, 2023 at 5:38 AM
Subject: Re: Web for Children Accessibility & WAI-Adap

Hi Janina, Sharon, Lionel and the Wai Adapt taskforce,


Thanks a lot for our meeting yesterday.

We will work on *use cases related to AAC imagery to the Web *in our next
meetings.
Would you have some *dates for our next meeting* to discuss the rest of the
cases and additional ones?

I did not want to interrupt the discussion yesterday as we where focused on
the case studies and it is at the core of the project.
Please find some considerations that might help if the WAI Adapt project is
meant to be AI -driven (even partly).
Do not hesitate to ask questions I just wanted to provide the information
so you can check if relevant.

Yesterday, I think Lionel mentioned a study. I assume it is the following
one, based on 12 papers, concluding trigger warnings cannot be used as a
mental health tool?
Complete Research A meta-analysis of the effects of trigger warnings,
content warnings and content notes <https://osf.io/qav9m/>.
In my understanding these trigger warnings are based on

   - conscious choices,
   - cognitive information
   - are not part of a more global trauma - based care approach.


It might be important to check effective behavioural techniques which
respect the best interest of all children (with specific needs)
The idea behind personalization, in many cases, is to change behaviours and
increase learning motivation.
This is why therapists have turned towards other semi- conscious strategies
as nudging (part of subliminal techniques) and is usually presented
positively in order to effectively change behavior.
"there are key principles in framing the choices that people are offered:
1) keep the message simple
2) if you want someone to do something, point out that lots of other people
are doing it
3) associate it with something positive rather than something negative."
*Well-being Science and Policy* - Richard Layard and Jan-Emmanuel De Neve
(Cambridge University Press 2023)

If I take the example of the triggers warnings:
-Changing a behavior through a completely conscious mechanism is very
difficult.
-Trying to change a behaviour through a negative message only, can also be
very ineffective.

Earlier this year, I worked on the AI audit framework for nudging on
children <https://arxiv.org/abs/2304.14338> by Marianna Ganapini
<https://arxiv.org/search/cs?searchtype=author&query=Ganapini%2C+M>, Enrico
Panai <https://arxiv.org/search/cs?searchtype=author&query=Panai%2C+E>
supported by IBM -Tech Ethics Lab.
Through the research, I added mitigation strategies for children with
disabilities.
The idea is to reduce the risk of nudging that would have detrimental
effects (ie.: encouraging addiction and lack of sleep through nudging
techniques)
but instead mitigating risks and fostering nudging linked to positive
learning behaviors.

For many children, the executive functioning skills take more time to
mature. So when offered options, consciously or unconsciously, it can be
very difficult for children to measure the consequences of choices and
evaluate trade -offs.
Knowing that AI as in increasingly more educational projects, I think it is
important to keep monitoring what is at stake, especially for children with
specific needs.
Therefore, we always wanted to be very cautious with our Community Group in
chosing the case studies with our group and believe it is important to
liaise with the COGA taskforce and community group so that children and
adults always get options framed in their best interest, maximizing agency
and safety.
Bob Dolan does a great work liaising with the COGA community Group and
Suzanne and I are liaising with Lisa Seeman from the taskforce.

Have a very nice day,
Maud