-

Of course, hardly anybody takes the time to add these descriptions. Even on Mastodon, where the software makes it easy to add them and, in many cases, even prompts you to do so, it's still an effort. And to be honest, it's an effort that humans should not <br />have to make. That's precisely the kind of busywork that should be done by computers, only until now they haven't really been that good at it.

Tue, 11 Apr 2023 00:37:48 +0200

atomicpoet@mastodon.social

Good point from @mistercharlie. If there’s any argument for A.I. doing social good, it’s in image-to-text capabilities—as demonstrated by Midjourney.

This could be a game changer for accessibility, and getting detailed alt text descriptions.

The Fediverse could be on the cutting edge of this tech since it’s easier here than anywhere else to add alt text descriptions.

https://www.lifewire.com/why-midjourneys-new-image-to-text-generator-is-an-accessibility-home-run-7377344

@fediversenews

show all 9 comments

Tue, 11 Apr 2023 00:48:08 +0200

Resting Facebitch

monsoonrains@mastodon.social

@atomicpoet @mistercharlie @fediversenews 100%

Simulated intelligence is such a good grunt work servant. It's already making mind numbing but necessary tasks easier.

Just a pity people are hyping it up to more than it actually is.

Tue, 11 Apr 2023 00:53:32 +0200

alice

lexd0g@wetdry.world

picture of a PSP battery inside a PSP in clip interrogator, the description says "a person holding a sony gameboy battery pack in their hand, with the battery back and the battery back, Aquirax Uno, nostalgic, a microscopic photo, panfuturism"

@atomicpoet @mistercharlie @fediversenews lol, i was just trying to use clip interrogator to see how well itd work for automatic alt texting, tl;dr not well, but its also made more so for ai art not image description and i also dont think it can read text

Tue, 11 Apr 2023 00:58:11 +0200

David Erik Nelson🪬

dave0@a2mi.social

@atomicpoet @mistercharlie @fediversenews I’m also seeing really promising stuff with AI transcription. All sorts of stuff more accessible to hearing impaired folks, language learners, folks with sensory processing issues, etc. is a solid win, in my humble.

Tue, 11 Apr 2023 01:05:52 +0200

Jupiter Rowland

jupiter_rowland@hub.netzgemeinde.eu

I myself wouldn't trust any AI with writing alt-texts for the pictures I use in my posts. These are almost exclusively pictures taken in virtual worlds.

First of all, an AI may be halfway capable of recognising virtual objects with exist in the real world just the same, but not perfectly so. As for objects which are exclusive to this kind of virtual worlds or even this particular world, objects which do not exist in real life, an AI won't be able to identify them.

Also, sometimes it matters where a picture was taken. An AI may be able to recognise well-known real-life places, but obscure virtual places, not so much, even less so if there are several such places which look almost or completely identical.

Lastly, even if an AI appeared that is fully capable of correctly describing what can be seen in pictures from virtual worlds, and by that I mean down to even automatically and accurately recognising the creators of objects in case this should matter, even then, the AI would not necessarily be able to know the context and thus decide what is important to mention, what is essential and what is unnecessary. Depending on the context, the very same picture can require different alt-texts, that is, unless people come and demand fully-detailed descriptions of everything that can be seen in a picture, for maybe it could matter to someone.

In case you've got enough patience to read a 6,000+-word rant, I've written a post about alt-text on pictures from within virtual worlds just recently. You can read it here.

Tue, 11 Apr 2023 02:02:09 +0200

FeralRobots

FeralRobots@mastodon.social

@atomicpoet
It would be TERRIBLE for accessibility.
Good alt-text is not strictly about describing the image - it's about SITUATING the image.
@mistercharlie @fediversenews

Tue, 11 Apr 2023 03:09:48 +0200

Olav

olavf@spacey.space

@atomicpoet
@mistercharlie @fediversenews as a prompt maybe. Anyone that lets LLMs write descriptions w/o editing is a fool.

Tue, 11 Apr 2023 06:39:01 +0200

Darby Lines

angry_drunk@union.place

@dave0 @atomicpoet @mistercharlie @fediversenews OpenAI Whisper has finally tipped the scales for my company to caption our training videos.

Tue, 11 Apr 2023 06:39:13 +0200

Eric Jennings

patamystic@sfba.social

@atomicpoet @mistercharlie @fediversenews

i let AI generate alt text for an image. this is what i got:

g/i« : L : 2t . = G = Z H i A ;i P e o x‘ i o . = = – AET 2 b 35 – . o S Y Z Z \ S 3 “A e e G e T TaE i & b T ‘ 2 Gl 0 > / b o g ko {1/ 3 B % e 2 o T biistie e ST R S f.‘ﬁ::% (! S A e A m 3 2 i e g v g R . At ,t‘ o et il R S B e T ; e s %‘ A 7 . Sty T j E NS . s = – o / By e i i = Ee ¢/ ! i . e S 4 £ 0 i 3 T s ey & v N i T E o : o b i : e e, 000 Lo o = L oo e s i (i B W e 7 7 1 kB v‘/ e R L v i G L LN e e . s . e % s S B G s o o oA ¢ i . A e e e i & GRS i 2% s SRR B N

the image can be seen here:
https://frightenthehorses.com/somebody-has-to-pay-for-easter-it-might-as-well-be-him/

Tue, 11 Apr 2023 06:46:45 +0200

Alexander Harding

alex@social.harding.dev

@dave0 @atomicpoet @mistercharlie @fediversenews I’d really like to see a #PeerTube plugin for #Whisper!