Full disclaimer: Before Jakob Nielsen’s very controversial article came out, I didn’t know who he was. I’ve been working intensively on the topic of accessibility for a few years now, but this name hadn’t appeared in my bubble until then.
Probably because I tend to follow the “new guard”, such as Sara Soueidan, Heydon Pickering, Adrian Roselli, Stephanie Eckles and others whose names I can’t think of right now.
I am not an expert in this field either, but through my work I have the opportunity to compare Jakob’s statements with my own experiences. There are already MANY reactions to this article, which Adrian Roselli has summarized here: Jakob has Jumped the Shark.
It wasn’t my plan to write about it, but after my wife read the article and had more questions than answers, I felt inspired to add my two cents. I chose the statements that raised the most questions for me. Here is the link to the article: Accessibility Has Failed: Try Generative UI = Individualized UX (archived).
Disclaimer
I use the term “LLM”, which stands for Large Language Model. I refuse to use to term “AI”, which stands for Artificial Intelligence.
While I can see the benefits for many other sectors and also as a sparring partner for writing code, these models haven’t shown any sign of intelligence regarding web accessibility. At least in my experience.
Summary & opinion
Out-of-touch, influential old guy working in web accessibility for many years claims that all hope is gone and only AI™ can save us. People who he thinks are disabled will receive a more “concise” UI, because he thinks they are not able to use a computer and assistive technology.
I think this is another big move by AI™ companies to buy community leaders and to make sure - once the European Accessibility Act is in action - that people buy their shitty “AI” powered overlays and don’t waste any time and especially money on actually improving the user experience.
And what really annoys me, is that people who are new to web accessibility might think that this guy is right and has a couple of “interesting ideas”. It sends a dangerous signal.
The statements
Accessibility has failed as a way to make computers usable for disabled users.
I tend to disagree. While the web is in a horrible state regarding accessibility, disabled users still seem to make it work and get around. There’s plenty of hardware and software choices to consume content on the web in whatever way works best for you.
Computers are still difficult, slow, and unpleasant for disabled users, despite about 30 years of trying.
Hard to say, do you know ALL disabled users out there, Jakob? Seems like a pretty general statement. If it wasn’t for many people trying, it would probably even worse.
Where I have always differed from the accessibility movement is that I consider users with disabilities to be simply users.
Huh, so all people are the same for you? Very honourable, but it gets tricky when you realise that people have different needs and cultural backgrounds, so you MIGHT want to reconsider as this approach could exclude a large chunk of folks.
Accessibility is too expensive for most companies to be able to afford everything that’s needed with the current, clumsy implementation.
This is true if it’s an afterthought on your agenda and not baked in to your development cycle to begin with. But shifting left can help to reduce costs while your whole team gets educated and more efficient during time, avoiding accessibility related costs in the long run.
Accessibility is doomed to create a substandard user experience, no matter how much a company invests, particularly for blind users who are given a linear (one-dimensional) auditory user interface to represent the two-dimensional graphical user interface (GUI) designed for most users.
Can you back that claim up? Can you define a “substandard user experience”?
By “old” users, I mainly mean people older than 75 years who start to exhibit major aging symptoms, such as weakened memory. These users need simplified navigation, simplified comparison features that do not require retaining information in short-term memory, and simplified explanations.
Sounds reasonable if you give them the choice to choose between a simplified (whatever that means) version and the regular (whatever that means) version. Otherwise it would be a pretty ableist move.
On the other hand, I don’t want to slow down a blind user with a screen reader blabbering through that word salad. Yes, I could — and should — edit ChatGPT’s ALT text to be shorter, but even after editing, a description of the appearance of an illustration won’t be useful for task performance.
Maybe leave it up to the user to skip your LLM-generated alt text? And how can you generally say it’s not useful? Some images can give context to a website or article and others are decorative. It always depends on the context.
Estimates from international reading research show that about 40% of the adult population in the United States can be classified as having low literacy. (This is about 100 million customers. Think dollar signs when you read these statistics.)
The sad truth is that most companies care about what’s profitable and not what’s ethically right or requested by a minority of users. However, I find this phrasing very weird coming from someone who should work with and for people. I have a feeling that you had dollar signs in your eyes when offered to write this questionable article.
We need an approach that scales, and that can support users with a wide range of conditions. Luckily, this is now emerging in the form of generative UI.
How about doing proper research, educating your team, implementing design with good ol’ HTML, CSS and JavaScript and sprinkle on some progressive enhancement?
If each department is in and aware, a little can go a long way here. But I guess it’s easier and more cost efficient to type into a LLM what website you want, copy the code and ship it to production? For a deep-dive, check out this article by Sophie Koonin: “AI”, and the trouble with inaccessible SaaS.
“Generative UI” is simply the application of artificial intelligence to automatically generate user interface designs, leveraging algorithms that can produce a variety of designs based on specified parameters or data inputs.
Based on the status quo of the web? Oh Jakob, then we’re in for a ride. 🚀 Cause most developers still don’t know how heading levels work, what an alternative text is or the difference between anchors and buttons.
In this second-generation generative UI, the user interface is generated afresh every time the user accesses the app. Most important, this means that different users will get drastically different designs. This is how we genuinely help disabled users.
How? I mean… how? What are the criteria for seeing different content? When do you decide that a user classifies as disabled? And why would disabled people want to see different designs than non-disabled people? So you decide who gets the full experience and who’s not capable of being able to enjoy it? Again, sounds pretty ableist to me.
For example, a more simplified experience can be shown to beginners, and advanced features surfaced for expert users.
Oh my god, will you stop throwing in vague fantasy ideas and explain HOW you would actually do it and WHAT the deciding factors are?!
We will no longer be designing the exact user interface that our users will see, since the UI will be different for each user and generated at runtime. Instead, UX designers will specify the rules and heuristics the AI uses to generate the UI.
This might sound good in theory, but do you think developers and designers will question the output of a LLM while their CEO is breathing in their neck and tells them to “move fast & break things”? It’s more likely they will just use whatever looks good and move on.
Besides creating optimized 1-D representations for blind users, generative UI can also optimize the user experience in other ways. Since it is slower to listen than to visually scan text, the version for blind users can be generated to be more concise.
Dude, this statement reminds me SO MUCH of the whole loudness war debate: “I don’t want to leave it up to the listener to turn up the volume, I’ll do it for them! The louder my record, the better it will sound than anyone else’s!”
My brother in Christ, have you ever considered in your 100+ years working in the field, that screen reader users can increase the speed of their app? That you don’t need to generate a more “concise” version but instead leave the choice to the users?