I wonder if ChatGPT or Deepseek could help me make website code that can
work with text-only browsers elinks or w3m.

On Fri, Jul 4, 2025 at 8:28 PM Eric Oyen via PLUG-discuss <
[email protected]> wrote:

> Um, welcome to my world. If the site isn’t setup logically with clearly
> defined areas, it can get rather confusing to me (yeah, totally blind and I
> hate dynamic content sites with a bloody passion). And yes, like that AI,
> if I can’t perceive placement visually, then it’s next to impossible for me
> to place the elements in a visually appealing style (I go with what sounds
> right given my ears and the screen reader).
>
> Btw, because of my situation, I like sites that use dedicated frames for
> the top banner, main content and footer items (one can add more and link
> them into the main content frame like many of thee built-in web servers on
> commodity wifi-routers).
>
> -Eric
> From the Central Offices of the Technomage Guild, the seeing isn’t
> believing Dept.
>
>
> On Jul 4, 2025, at 2:36 PM, David Schwartz via PLUG-discuss <
> [email protected]> wrote:
>
> I spent about 5 hours last night working with Claude AI to help build a
> web page where it was helping me update a bunch of html/php and css files
> that it generated itself. The two main pages were a lot fancier than I
> would have made, but it had a lot of trouble making them look right in all
> situations. I had it generate a footer with standard pages (Privacy, Terms,
> About Us, etc.) and while they were all the same layout, I had to go
> through nearly 2 dozen iterations to make them look decent on all platforms!
>
> I was amazed by the css this thing generated, particularly because I’m
> horrible with CSS. But I can’t believe how many times we had to keep going
> over the same things because of one change after another that did not work.
>
> If it’s just logic that can be clearly tested, that’s one thing.
>
> But AI cannot “see” things; and while we humans can, it’s hard for us to
> explain visual issues to a mechanism that’s basically blind and cannot even
> display what it’s doing on the screen.
>
> I was viewing this page in several browsers on my desktop, a large iPad,
> an iPad Mini, and my phone. I’ve gotta say, this crap drives me nuts! I had
> to keep suggesting ways to fix it and it would sometimes take four or five
> attempts for it to get it right! I’m no good with css, but I can tell
> immediately when a change has not altered how the screen looks when it
> should have changed. The fact that it can’t only makes it go a LOT slower.
> It would have taken a lot longer for AI to just cycle through a bunch of
> guesses to fix this if I wasn’t able to guide it to some alternative
> solutions. I feel bad for people with no tech background relating to UX/UI
> stuff who have to rely on AI to do these things.
>
> The future of UI design assistance with AI is going to have to be done
> with a WYSIWYG drag-n-drop interface. Maybe that means having AI build it
> initially and then using a normal web page editor to make changes to the
> UI.
>
> Has anybody here had similar experiences and found a good solution for UI
> editing with AI’s help?
>
> (I have always hated the “look-and-feel” side of software. This is exactly
> why.)
>
> -David Schwartz
>
>
>
>
> ---------------------------------------------------
> PLUG-discuss mailing list: [email protected]
> To subscribe, unsubscribe, or to change your mail settings:
> https://lists.phxlinux.org/mailman/listinfo/plug-discuss
>
> ---------------------------------------------------
> PLUG-discuss mailing list: [email protected]
> To subscribe, unsubscribe, or to change your mail settings:
> https://lists.phxlinux.org/mailman/listinfo/plug-discuss
>
---------------------------------------------------
PLUG-discuss mailing list: [email protected]
To subscribe, unsubscribe, or to change your mail settings:
https://lists.phxlinux.org/mailman/listinfo/plug-discuss

Reply via email to