TechTakes

1491 readers

95 users here now

Big brain tech dude got yet another clueless take over at HackerNews etc? Here's the place to vent. Orange site, VC foolishness, all welcome.

This is not debate club. Unless it’s amusing debate.

For actually-good tech, you want our NotAwfulTech community

founded 2 years ago

MODERATORS

[email protected]

138

Neil Gaiman on spicy autocomplete (www.tumblr.com)

submitted 7 months ago by [email protected] to c/[email protected]

13 comments fedilink hide all child comments

Many magazines have closed their submission portals because people thought they could send in AI-written stories.

For years I would tell people who wanted to be writers that the only way to be a writer was to write your own stories because elves would not come in the night and do it for you.

With AI, drunk plagiaristic elves who cannot actually write and would not know an idea or a sentence if it bit their little elvish arses will actually turn up and write something unpublishable for you. This is not a good thing.

top 13 comments

sorted by: hot top controversial new old

[–] [email protected] 32 points 7 months ago (3 children)

the splash damage this stuff is causing to the cultural commons is going to be extremely hard to quantify, and undo

I've been pondering it for a little while now and even just spitballing on axes of bitrot/vc-exhaustion/etc, we're going to have all this useless space-filling shit on our collective doorsteps for 5~15y

there appears to be some amount of hope ito things that auto-blacklist sites, etc. but since we don't have real transclusion (and other source-strong models of hosting/serving), provenance problems are going to remain at a nasty prominence for quite a while too

and the fucking shitheads that push this may never be brought to account -_-

[–] [email protected] 17 points 7 months ago* (last edited 1 month ago) (2 children)

It’s already too late for a lot of places, imo. DeviantArt for example is overrun by LLM-generated sludge and no amount of cleanup will undo that; and that site has been a staple of amateur and upcoming artists for decades. The same seems to be happening to Pixiv (which is big in Japan), too. Search engines are also full of generated SEO spam and it’s getting worse, with image search being close to useless unless you do implement some sort of blocklist. Which, for that use case, luckily already exist and aren’t bad (shameless self-plug), but it’s still a manual step you have to take and won’t help my grandma who’s looking for cookie recipes.

The silver lining might be that a growing number of people are willing to try decentralized solutions. I’ve seen more non-techies come over to Lemmy, Mastodon and Misskey as a result, but it’s still sad to see, especially because this will ultimately lead to tons of older content becoming either lost or needles in a shitstack you can’t ever hope to recover.

[–] [email protected] 12 points 7 months ago

There are a few art sites that swiftly banned AI art before they got flooded (with Newgrounds and FurAffinity the most well-known).

Alongide the decentralised solutions you mentioned earlier, those sites should also come out okay once the dust settles.

[–] [email protected] 10 points 7 months ago (1 children)

(shameless self-plug)

nice. I've got one of the others after I found it a while back, but it looks like yours has a narrower scope?

for places like deviantart... yeah. I don't really know how you reverse that. containment by clearly marking all the profiles/uploaders (after some way it was detected)? community maintained lists which help defend?

it's such a fucking mess. and we have no choice but to deal with it, otherwise it will just make things even worse. that's the thing that really fucks me off about this stuff. all the undue extra work created for many, many people by the inconsiderate actions of far fewer that nonetheless don't care about their actions.

[–] [email protected] 7 points 7 months ago* (last edited 7 months ago)

nice. I've got one of the others after I found it a while back, but it looks like yours has a narrower scope?

Yeah, that one actually includes mine and covers more bases. When I started my list, I think they only had a uBlock-based blocker, which was too aggressive for me (and does not work properly on mobile), and there were many small uBlacklist lists which I just combined into one.

I think these days most of them are similar anyway.

[–] [email protected] 7 points 7 months ago (1 children)

and on that topic, some actual numbers (posting without reading - that'll happen as soon as my delivery gets here)

[–] [email protected] 5 points 7 months ago* (last edited 7 months ago) (2 children)

something I don't really understand is why wikipedia doesn't have an archiving policy for links. actually I don't really understand why wikipedia isn't effectively building a second wayback machine

[–] [email protected] 9 points 7 months ago (1 children)

it does! Every reference link gets submitted to the IA

[–] [email protected] 4 points 7 months ago (1 children)

Is that a recent change? I feel like I've seen a fair number of dead links on Wikipedia, mainly on older unpopular articles

[–] [email protected] 5 points 7 months ago

The archives aren't automatically added to the reference text in the articles. There's an IABot you can feed particular articles to that will add it to reference templates. Also, if the IA archiving fails to work, too bad. (e.g. how it can't even archive Twitter any more.)

[–] [email protected] 3 points 7 months ago

yeah it is rather a strange shortcoming/oversight

[–] [email protected] 2 points 7 months ago

Because it was done with and for money. And the criminals that are supposed to legislate about this can't wrap their brains about anything other than the money involved with it.

[–] [email protected] 7 points 7 months ago

Faunits to the rescue!