I also agree that we shouldn't be creating separate experiences for the blind. I think it's generally acknowledged that they end up being worse than a combined interface, never getting the resources or new features that experiences for sighted users get.
Where we seem to disagree is on the roll that screen readers play in limiting the usability of technology. On the one hand they are amazing because they provide access to technology that would otherwise not exist. On the other hand, by virtue of the way they function-mapping a 2 dimensional visual experience into a one dimensional stream of audio-using a screen reader can only be so efficient.
This lack of usability puts access to technology beyond the reach of many who are less tech savvy than you or I, and given that the vast majority of people losing their vision in the US are the elderly, there are a lot of people who fall into that category. What's worse, the rate of vision loss is set to double as baby-boomer's age out.
I totally agree that the medical-model of Accessibility sucks, but I think Screen Readers fall into that category. They seek to adapt an experience designed for others to the needs of the disabled. Conversational interfaces have the potential to create a consumer quality experience, that by it's very nature is accessible (at least to the blind). And accessible by default is the best possible outcome.
Anyhow, I look forward to reading more about your SDK. Where can I learn more? I'm building an app that could benefit from a conversational UI on top of the traditional one and would be interested in reading up on what you offer, particularly as it's meant for blind users too.
One thing screen readers are super good at is exposing shitty IA design, which is regrettably common.
That said, it cuts both ways. There is a public transport app in the UK (Traveline GB) that as a low viz (legally blind) user I find incredibly frustrating to use, but my no viz pals absolutely love.
In this case it seems the IA is there but the visual interface to it is worse than what voiceover exposes.
Accessibity is hard.
That's a very low bar. Expert computer users like you and me have no problem with a screen that has even thousands of pieces of information on it. We use contextual cues to help us navigate that space, relying on good information design. But most people find current computers overly complicated. I think one of the reasons smartphones have eclipsed PCs is that the 4 inch screen forces you to cut down to a more human scale of choice complexity.
But that's why I am finding myself agreeing with OP that conversational interfaces are the future, for the same reason. They will force us to design down even further, into a one-dimensional UI. Choice, choice, choice, choice, etc.
That UI will be a slight simplification on the current phone UIs, and of course multiple threads can be shown on the screen at once to allow better use of screen real-estate for those who need it. Essentially, you just add lots of these one-dimensional UIs to the screen and multi-task. But the capital-T Truth is that even without a screen reader, using a Desktop computer and indeed vision itself is a largely linear process. Experts get fast at it, but we still must scan about to discover new information and places to interact. I think we will be surprised how much of our computational life actually becomes conversational.
One of the most exciting things to me about conversational interfaces is that writing automation software that uses them is much more trivial than writing automation software for a big complex GUI. And building AIs that live in them is also easier.
In general, I think there is something very important in the space of UI for beginners, mobile users, AIs, and people with disabilities that is happening. A kind of crystalization of intent. I'm only hearing people like OP start to talk about it now, but I actually think this is the next big wave in UI. And I do agree with you that most time spent improving UIs is wasted in exactly the way you describe.
So yes, I generally agree that conversational interfaces are a good direction, and I'd like to incorporate them into the app I'm currently building. I just worry that the tone of this article makes the current state of the art sound terrible, when it's less the state of the art and more the implementation details of that state that makes things difficult.
If you build conversation into the core of your UI framework in an intrinsic way, it means everything written in that framework will be done right. Every 2D interaction has a first class conversational analog because conversation is the base interaction primitive and all of the 2D interactions exist within that space.
Are screen readers difficult to use? Sure, but no more so than typical desktop environments. Describing them as "requiring special training from organizations like the Lighthouse for the Blind" feels like a bit of a stretch. Would some folks require training? Sure, just as they might require training to learn OS X if they're familiar with Windows, but that's true of anyone sighted or not, and hardly a requirement. Also, while I agree that licensing costs for some screen readers are enormous, just about every mobile and desktop platform has at least one free, competent alternative. Someone newly blind or blind after years of sighted life might certainly need training, but in general I find picking up a new screen reader to be no different than learning any other semi-complex app. If using Word/LibreOffice is part of your job then you sit down, learn it and become competent with it over time, or you get trained. Categorizing screen readers as being different and worse in this regard feels unfair.
"All of the time and energy that goes into creating the perfect user experience for an app is wasted, or even worse, adversely impacting the experience for blind users." Again, I strongly disagree with this. The problem is that UX time and energy is misspent almost exclusively on visual aesthetics, or even worse, on a certain type of visual aesthetic. For instance, say you use icon fonts because they're a cool thing to do. That choice has accessibility impact beyond blindness. Had a UI/UX designer focused on universal design and on crafting experiences that both look well and perform nicely from a keyboard/switch user's perspective, that interface will be visually appealing, nice to use via the keyboard, easy for users with low hand dexterity using switch controllers, etc.
Again, kudos to you for driving innovation in this space. I wish access tech didn't always feel like an industry stuck a few decades in the past, or wasn't being driven forward by companies with a medical-model concept of disability who view us as a market to be pitied rather than sold to. To that end, I like seeing exciting tech in this space, even if I can't imagine ever preferring to shop by a conversational interface. But I read headlines like "Apps are a nightmare for the visually impaired" and think "huh, I'm building an app for the visually impaired and hope it isn't a nightmare." And I worry that, rather than learning about how universal design can build experiences that look and work well for everyone, software developers and companies will read this post and think "I don't have to make my app accessible because blind people are using special custom interfaces anyway, so when a blind person tells me my app isn't accessible, I can pretend ARIA attributes are hard and write them off." Again, not trying to be negative, I just wish folks understood that we don't need separate interfaces and experiences.