Post by @kimcrawley@zeroes.ca

RE: https://sfba.social/@drahardja/117136649933380203

Same.

I in fact block every single shithead using AI slop anywhere.

It accelerates environmental destruction, practically tortures the poor people who are forced to live near the datacentres, and emboldens technofascism.

There are literally no excuses for posting or sharing AI slop images, text, anything. And I will instablock you for as much as sharing a post with AI slop.

Dave Rahardja

@drahardja@sfba.social 19 hr. ago

Hi blog writers: You don’t have to attach a unique image to every blog post, and you *certainly* don’t need to use #AI to generate an image for every blog post. You may think AI-generated images make your posts more attractive, but you are mistaken; they are doing the opposite: they make your posts more *repulsive* to a good number of potential readers. If I see an AI-generated image in the preview, I won’t click through to read your post, I won’t boost the post that mentions your post, and I certainly won’t read the rest of your blog.

5 Replies

0 Boosts

Elena Brescacin

@elettrona@poliversity.it 1 hr. ago

@kimcrawley @drahardja I am blind. At the beginning of this AI trend I was happy about image generation and even got the illusion "I can create visuals" but then it has no control. I usually interact with an AI trained for blind, it's called "be my AI", to describe images. So I could generate pics, then have them described. But it can easily allucinate. If I generate images, I have no full control on them anyways, so I do what I can do: WRITE.

I use however AI coding feature when I have to satisfy some needs that aren't covered by free or even paid software: too many "traditional" developers, especially in the linux environment, DO NOT THINK of accessibility; many collaboration systems such as OnlyOffice for example, despite statements, have very poor accessibility issues and most answers are "do it yourself if you need" so, for building custom tools, AI has increased self-determination for us.

I know many vibe-coding blind folks creating and sharing software, some very good, other very poor. But I work in different way - aware of how much dirty code there is, and how many vulnerabilities I could spread, I play on my own covering my needs, and stop.

The exception for images though, comes with Castopod: I could build a podcast, audio version of my written contents, or even some audio demos, and federate it with Castopod. But guess what? For RSS reasons, you MUST include a 1400x1400 cover image. Even if the visual content is completely useless for readers/listeners, the paradox is that I could independently do whatever with audio, but the mandatory picture prevents me from even creating a podcast! So? I would need a sighted person just for image or... There could be AI generator.
But, fanatic folks like you, MANIPULATING us with guilt, say "if you generate image content with AI I wouldn't even read", are creating a barrier to people using AI as an assistant for disability. Like saying "don't enter with your guide dog, I'm afraid of dogs".
Environmental problems exist. Dangerous authoritarian politicians and tech leaders exist. Privacy issues as well.
But, what fucking alternatives do we have as blind people? Who on Fediverse has ever been worried of "let's create a 24 hours available human-driven worldwide photo description network?" No one. Be My Eyes has videocall volunteering (centralized) since 2014, BeMyAI exists since 2023, but humans (SIGHTED humans) are always ready to criticize and place boundaries, without proposing real alternatives for us, allowing the mentioned tech leaders with no regards for humanity and environment, to make us (vulnerable people) dependant on them.
Too many "no AI" movement folks are worse than taliban.
Am I a taliban pro-AI? No. I know its limits, its dangers, but when it can work as an aid, I do not want folks around behaving like my parents did when I was a child "eat, baby. Empty that bowl! Think of African kids of your age, dying for hunger!" F. Off. Heartfelt.

Dave Rahardja

@drahardja@sfba.social 1 hr. ago

@elettrona @kimcrawley The use of AI for accessibility is fine with me, because as you wrote, people don’t usually think about accessibility in their products. At the same time, the problem of AI slop also exists and it’s literally ruining our world. How do we extract the good without allowing the bad?

Elena Brescacin

@elettrona@poliversity.it 1 hr. ago

@drahardja @kimcrawley It's what I'm thinking and I'd like so much to discuss it with other people (blind people, but also Fediverse administrators etc) because limiting damages of AI slop is our duty - on audio, on video, on coding.
Prohibiting electronic voices implies preventing people who hear but have speech problems, to create something to listen; a mandatory AI-generated cover image could be generated by a blind, so, what could we do?
It would be a humiliation forcing to say "I was a podcaster but I cloned my voice because a cancer made my vocal chords unusable" or "I generated my cover image with AI because I'm blind".
Except for very specific topics, someone shouldn't be obliged to disclose their health or disability condition; but how to avoid bad-quality contents preserving people with particular needs? This debate will be very long and I'm determined to bring it wherever it counts - at least in environments fighting AI slop.
Slop must be pushed out from web, but we must learn to distinguish a prothesis from a completely artificial content, created by a mindless lazy "normie" with full wallet.

Dave Rahardja

@drahardja@sfba.social 1 hr. ago

@elettrona @kimcrawley I also think there is a difference between using AI (ML, classifiers, transformers) to perceive the world, and using AI to generate content. I’m much more accepting of the former, and much more critical of the latter.

Elena Brescacin

@elettrona@poliversity.it 56 min. ago

@drahardja @kimcrawley I agree. Generating content isn't appropriate especially if you have a blog you claim as completely yours.
The podcast covers I was talking about, would be a SINGLE image. The same, which is required by podcast platforms to accept compatibility via their rss feeds. So Castopod has made it mandatory.
But, for what concerns me at least, when I write, I don't place images because I am simply unable to have full control on them so there could even be the best generator, but it would always be a barrier with high risk to create trash.

The only occasion I published a content generated by AI was when my best friend turned 50 years old. Inspired by a far-right propaganda articles talking about a very artificially touching child-related story, I prompted GPT to write a content, to demonstrate to my reader how much an artificial article could be near to reality if you can properly prompt the robot.

1 more reply