Email or username:

Password:

Forgot your password?
All posts Mozilla's posts Post Back to profile
Top-level
d@nny "disc@" mcClanahan

@sasha92 @mozilla and uhhh clearly hannah gives a shit which is why she built the distributed alt text database. mozilla clearly isn't doing its job as an org so individual hackers on the internet are doing it for them and that's obviously much harder. i spelled out how they can do better, you're giving them a pass because they did something cool once. guess which one of us cares more about mozilla's success?

3 June at 11:01 | Wall-to-wall | Open on circumstances.run
15 comments
Emi

@hipsterelectron @sasha92 @mozilla There is billions of images online and an external human can't describe each picture that's missing alt text.
That project has a different goal than Mozilla's alt-text AI and I am sure you can use both - human descriptions with that project for the few images that will have it and Mozilla's AI for the rest.

3 June at 11:29 | Open on tilde.zone
d@nny "disc@" mcClanahan

@paper @sasha92 @mozilla you do know there are billions of humans right? wikipedia is widely used to train statistical models but it too is the compilation of many humans over time. adding alt text to an image takes about as much effort as editing a page (and often much less). do the math

3 June at 11:34 | Open on circumstances.run
d@nny "disc@" mcClanahan

@paper @sasha92 @mozilla also, as noted above, it is generally accepted by experts on alt text (not experts on "AI", who have a conflict of interest) that it cannot be machine generated in any remotely meaningful way. this is like providing an escalator to wheelchair users

3 June at 11:37 | Open on circumstances.run
Emi

@hipsterelectron @sasha92 @mozilla There are 3.2B images uploaded in a day (1.2T /year), many of them are repeating, google has 130B indexed. You can't describe all of that. Sure, human description will probably be better in many cases, but AI descriptions are still very useful.

Also, I doubt that project will get as many people editing it as Wikipedia has, so it can be great for a few popular images, memes, etc. but it can never cover random images on social media and websites without alt text.

3 June at 11:54 | Open on tilde.zone
d@nny "disc@" mcClanahan

@paper @sasha92 @mozilla if you're not blind or visually impaired you absolutely don't get to make that call (and single individuals do not represent an entire community). hannah's project was built in direct conversation with the disabled people it serves. you're finding reasons to trash it to justify your preconceived notions and i think that's a really terrible thing to do. blocking now

3 June at 12:01 | Open on circumstances.run
drevil

@hipsterelectron @paper @sasha92 @mozilla nobody is trashing on anything bud, they are just saying that it would be physically impossible for any group of people, regardless of how large they are, to add a description to a fraction of the imagines uploaded every day.

You could, instead, argue that it would be nice to use image detection to match images with multiple databases that contain descriptions, so if you don't feel like using the model (and/or you don't like what it gives you) you can try to check a community provided one, if any exist.

@hipsterelectron @paper @sasha92 @mozilla nobody is trashing on anything bud, they are just saying that it would be physically impossible for any group of people, regardless of how large they are, to add a description to a fraction of the imagines uploaded every day.

You could, instead, argue that it would be nice to use image detection to match images with multiple databases that contain descriptions, so if you don't feel like using the model (and/or you don't like what it gives you) you can try...

3 June at 12:39 | Open on tilde.zone
Emi replied to drevil

@chickfilla @hipsterelectron @sasha92 @mozilla I can't reply to them as they blocked me, but I just wanted to say that I do see the value in that project and I like what they are doing. I am just saying that while this project will not give as accurate alt texts as that human made database, I think it is still valuable because of how many images without alt text exist out there.

3 June at 14:06 | Open on tilde.zone
drevil replied to Emi

@paper @hipsterelectron @sasha92 @mozilla Oh and I agree with you entirely, I wrote another reply where I basically said the same thing. I think it is a very noble and useful cause, my issue was with the nature of their replies instead.

3 June at 14:25 | Open on tilde.zone
d@nny "disc@" mcClanahan replied to drevil

@chickfilla @paper@tilde.zone @sasha92@c.im @mozilla please read up on hannah's work before suggesting things like you're the first person to ever consider this problem. "image detection" like you describe is totally separate from "AI" bullshit (so why are you even mentioning it like it's some sort of gotcha?) and is very much a component of hannah's existing implementation

3 June at 14:15 | Open on circumstances.run
drevil replied to d@nny "disc@" mcClanahan

@hipsterelectron @mozilla I think you misunderstood what I am trying to say. I am suggesting that one could use it to relate an image with community supplied alt texts should you want that as an option and they be available.

Also I am not looking for any gotchas, I do think a project focused on providing image descriptions is a noble cause, my problem comes with your idea that a model has no use for this. If this creates an image description that the poster likes and feels like it represents the image they are about to post well, then what's the harm?

You are acting extremely defensive when nobody, and I mean nobody, is discrediting nor even criticizing their project, just pointing out obvious limitations that do not in any way discredit its intentions.

@hipsterelectron @mozilla I think you misunderstood what I am trying to say. I am suggesting that one could use it to relate an image with community supplied alt texts should you want that as an option and they be available.

Also I am not looking for any gotchas, I do think a project focused on providing image descriptions is a noble cause, my problem comes with your idea that a model has no use for this. If this creates an image description that the poster likes and feels like it represents the...

3 June at 14:24 | Open on tilde.zone
d@nny "disc@" mcClanahan replied to drevil

@chickfilla ok then i have no clue what you're trying to say. are you just trying to paint me as defensive? sorry for assuming you were acting in good faith

3 June at 14:26 | Open on circumstances.run
sasha

@hipsterelectron @paper @mozilla What is "an expert on alt text"? Is that a linguist?

3 June at 12:04 | Open on c.im
Yusuf Toropov

@hipsterelectron @paper @sasha92 @mozilla "generally accepted by experts on alt text" that this can't actually work sounds a bit weasel-languagey, sorry. I'm not an expert on anything and I'm going to use this. That's if it's okay with @hipsterelectron , I mean.

Not every tool incorporating AI is evil. Some of it is both useful and transparent. AT THE SAME TIME. Mind boggling.

3 June at 14:48 | Open on toot.community
sasha

@hipsterelectron @mozilla I didn't comment on the merits of the project you propose they adopt. Saying "do exactly as I say" or you don't give a shit is not the most productive approach, to say the least.

3 June at 12:04 | Open on c.im
d@nny "disc@" mcClanahan

@sasha92 @mozilla it's called constructive criticism actually you should try it sometime

3 June at 12:11 | Open on circumstances.run
Go Up