Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Should AI editable, Identify autogenerated text, and algorithm bias be included in WCAG 3? #110

Open
rachaelbradley opened this issue Aug 22, 2024 · 0 comments

Comments

@rachaelbradley
Copy link
Contributor

I really applaud the direction of travel WCAG 3.0 is taking. It reminds me of gray-box testing; essentially a testing method that falls between white box and black box testing to help identify defects in applications. In gray-box testing, as you may know, testers can access the rendered UI and have proxy information about the system's internal workings (in this case, perhaps provided through the DOM or Accessibility Tree), but they cannot see the actual build code of the component being tested. The major benefit from this approach is that a test team can undertake an assessment without the need to question developers - so the assessment is discrete.

Almost all of the WCAG 3.0 Outcomes appear to be specified in terms of a gray-box testing approach, for example, "Audio shifting designed to create a perception of motion is avoided; or can be paused or prevented”. I say almost all as I have spotted a couple of outcomes which break with this concept:

  1. AI editable: Auto generated text descriptions are editable by content creator.
  2. Identify autogenerated text: Auto generated text alternatives are identified.

The issue with both of these is that from looking at the DOM or Accessibility Tree alone a test team could not say if text descriptions have been generated; and would not be able to say if a content creator was able to edit them. They would be forced to talk with the developers.

Thinking about these two outcomes, I would also question why the use of AI text descriptions is an accessibility issue in itself; when surely it is only the quality of the text description which is impacting (and nicely caught under other outcomes).

In light of the above, are these two specific outcomes in fact necessary to include in WCAG 3.0?


I missed one - Algorithm bias: Algorithms (including AI) used are not biased against people with disabilities.

p.s. How a test team would even go about testing this one accurately, I don’t know!

Issue submitted via public-agwg-comments email. Respond to email with update when addressed.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant