Blind internet users struggle with error-prone AI aids(ft.com)
ft.com
Blind internet users struggle with error-prone AI aids
https://www.ft.com/content/3c877c55-b698-43da-a222-8ae183f53078
23 comments
When I say I don't deal with overlays, and I refuse, I think they're a scourge. They want to help businesses by not getting sued, but they don't want to help us, the community, in any way. Audio eye threatened to sue someone who was in the process of exposing them, for defamation.
https://adrianroselli.com/2023/02/audioeye-will-get-you-sued...
Oh I also don't deal with overlays. I have an extension installed to block all of them.
> The club had “no idea” the software, set up with no manual intervention, was that unreliable, Rosin said.
It would be comical if it weren't so sad that software were making lives worse, not better.
It would be comical if it weren't so sad that software were making lives worse, not better.
It's maddening that they didn't check if it worked. It's like they don't care...
It's a thing I run into with a decent amount developers these days, actually. I tried to work with a particular service to get them to fix their accessibility for months, and all I got was well, we fixed the tabbed navigation a bit. Which would probably be cool, if this wasn't a webapp that I was trying to use with all my various quick nav things, etc.
Well, it's likely that they don't.
The rise in AI generated content affects us all, generally negatively, sighted or not.
There are a vast number of sites that are largely LLM generated that exist to deliver adverts. Basically, the old school link farms have evolved into automatically generated content farms.
For example, if you search for a few IT technical terms you will often get a lot of blog style returns. They will often look like a Word Press site with a selection from a colour and font scheme. With luck you will find some Stack overflow posts and with the Good Lord (pick your deity) looking down, you might get some real content.
The situation for finding and engaging on the internet is pretty dire and if one or more of your faculties are a bit shag ... really dire.
Perhaps what we need is some sort of Open Access to add to Open Source. I have deliberately picked Access even though its been sort of taken by https://en.wikipedia.org/wiki/Open_access
The Access I envisage is actual ability to access something. The other effort might consider Open Research or similar instead.
There are a vast number of sites that are largely LLM generated that exist to deliver adverts. Basically, the old school link farms have evolved into automatically generated content farms.
For example, if you search for a few IT technical terms you will often get a lot of blog style returns. They will often look like a Word Press site with a selection from a colour and font scheme. With luck you will find some Stack overflow posts and with the Good Lord (pick your deity) looking down, you might get some real content.
The situation for finding and engaging on the internet is pretty dire and if one or more of your faculties are a bit shag ... really dire.
Perhaps what we need is some sort of Open Access to add to Open Source. I have deliberately picked Access even though its been sort of taken by https://en.wikipedia.org/wiki/Open_access
The Access I envisage is actual ability to access something. The other effort might consider Open Research or similar instead.
I've already personally encountered a website in search results that when visited shows just a headline/title that is suspiciously spot-on to your search parameters, then a bunch of content-loading placeholders with ads interspersed. After a few seconds the placeholders start filling in, one section at a time, with what is clearly AI-generated filler text related to the title topic.
I don't know how worse it will get, but my current solution is a hoard of pdfs (books and articles), saving links (in case I can't find that one good site), and a documentation browser (Dash on macOS). A current trend is curated lists of links (awesome lists, blogrolls,…) so maybe one day, the only good way to experience the web will be this small web which will act like a human filter.
Do you hoard web pages using extension like SingleFile Web?
I probably should, but so far, there's only a few articles I've saved this way (long-form writing that get saved to my read-later list). Most of the links I've saved are objects-related (apps, libraries,...), and will only be useful if I need to interact with the object. If something is useful to me (techniques, recommendations,…) I put it in my notes. If the whole site is useful, I archive it.
Good idea to stash stuff away. I have access to an IT company (I own it) and a fair amount of storage but nothing like Archive!
One day we shall meet on HN and link our Nextcloud instances, links to some weird static HTML and so on and form a new web of information.
The internet (web) will never die, per se, but it does feel awfully siloed off at the moment.
Mind you I do manage to run my company email on prem (Exchange - soz, but HA Proxy fronts its webby stuff and Exim is my SMTP weapon of choice). I also run my family email domain and several others too. No hyper-scalers here, thank you very much.
I'm looking into the fediverse. I think that's where this stuff really belongs - the front pages of the internet have become mega corps. Its all looking a bit half hearted in fediland but Lemmy and Mastodon and co are starting to look quite useful.
The internet is still really diverse and flourishing but it needs a right good kicking as do the users of it.
One day we shall meet on HN and link our Nextcloud instances, links to some weird static HTML and so on and form a new web of information.
The internet (web) will never die, per se, but it does feel awfully siloed off at the moment.
Mind you I do manage to run my company email on prem (Exchange - soz, but HA Proxy fronts its webby stuff and Exim is my SMTP weapon of choice). I also run my family email domain and several others too. No hyper-scalers here, thank you very much.
I'm looking into the fediverse. I think that's where this stuff really belongs - the front pages of the internet have become mega corps. Its all looking a bit half hearted in fediland but Lemmy and Mastodon and co are starting to look quite useful.
The internet is still really diverse and flourishing but it needs a right good kicking as do the users of it.
The Fediverse is definitely where all the stuff belongs. I do use a service for all of this. I pay $12 per a year, and have stashed away quite a lot of stuff there.
Error-prone AI has been a horrible nuisance even for those who aren't blind.
“AI aids” is a wonderful accident of language.
It isn't something I struggle with, because it isn't something I choose to deal with at the moment.
Can confirm. But it's barely AI. Definitely not generative. Articles just love to put AI on everything. Then again, so do these accessibility grifters that plop the label onto anything that's barely automated.
This article is behind a paywall, I unpaywalled it for everyone.
https://archive.ph/kED3D
Who knew you could get aids from AI...
This news hit the accessibility community like a ton of bricks. Overlays have been seen as worse than useless among accessibility professionals, so it was a huge shock that a reputable provider would have acquired an overlay company. The structure of the transaction was also weird (the CEO of the target company purchased stock in the acquirer as part of the transaction).
Soon thereafter news broke that another accessibility rockstar was joining AudioEye (a company similar to UserWay). [1]
I imagine that the dam may be breaking and tech-enabled solutions may not be viewed in as bad a light as before.
1: https://www.forbes.com/sites/stevenaquino/2024/01/31/new-aud...