oh someone finally offered cory doctorow a big enough sack of cash to do a cory doctorow?

[scott@mastodon.clitheroe.ca]

@zzt I actually wondered if being accused of indulging in purity culture in that piece stems from a fear that when when they do, some substantial segment of his core audience simply won't need/want to read him anymore?

[zzt@mas.to]

@mos_8502 here’s a good example from an engineer I don’t follow https://bsky.app/profile/coloradotravis.bsky.social/post/3mezgaylny22p

I have more but unfortunately they’re engineers I do follow and would prefer not to link. one of them posted a Datalog love poem from an LLM as proof of something???? and I read it and the Datalog was incorrect

[mos_8502@studio8502.ca]

@zzt I have this vague feeling that there's some underlying need for "true" sapient AI that's feeding this.

[mos_8502@studio8502.ca]

@zzt Me personally, I'd love to have a sentient android friend, but reality just doesn't support that right now. Maybe one day far in the future someone will crack real AI, but these models are nowhere near real cognition.

[mos_8502@studio8502.ca]

@zzt I could interact with ChatGPT or Claude all day every day for a year and I would never begin to think it a sapient being. I know what it is, I know what it can and can't do. I can, subjectively speaking, feel the difference between talking to it and talking to a person, even if I can't articulate the difference well.

[zzt@mas.to]

@mos_8502 as an outside observer I just wanted to read a Datalog love poem from that person’s SO, because I think that’d be a really cool thing to receive

unfortunately I didn’t get that; I got some plagiarized trite horseshit poorly translated into broken Datalog by a machine learning model built to make shitty translations more convincing

what’s interesting in all of these cases is that the LLM output is very special to the person who generated it, and not at all to most others.

[joblakely@mastodon.social]

@aburka @zzt
I didn’t see any AI boosterism. I’m not sure I understand the issue in that article with the purity culture quote. It wasn’t referring to AI, AFAICT. It was referring just to being on bluesky.

[aburka@hachyderm.io]

@JoBlakely @zzt there were at least two remarks about purity but one was referring to LLMs

> Doubtless some of you are affronted by my modest use of an LLM. You think that LLMs are "fruits of the poisoned tree" and must be eschewed because they are saturated with the sin of their origins. I think this is a very bad take, the kind of rathole that purity culture always ends up in.

[aburka@hachyderm.io]

@mos_8502 @zzt you may be correct but many have thought this and been wrong

[zzt@mas.to]

oh good, the “you’re just doing purity culture” thing is already taking hold over on bluesky

so the line is now supposed to be that local LLMs are good and moral and SaaS LLMs are bad, when local LLMs come from the same fucked system that’s also actively making it impossible to buy computing hardware powerful enough to run even a shitty local LLM? is that about right? I’m supposed to clap cause someone with money is running a plagiarism machine but slower and shittier on their desktop?

[da_gut@dice.camp]

@zzt I thought that with local models you strictly controlled what information that they were drawing upon?

[zzt@mas.to]

@Da_Gut that’s incorrect for all of the local, supposedly open source models I know of

all of the research I’ve read on this has easily extracted verbatim plagiarized text from the models, because all of them have their origins in the same sources — usually Facebook’s leaked llama model or deepseek (which itself took from previous models). it isn’t possible for LLM models to be trained by anything other than a billion dollar company or a state operating like one.

[orb2069@mastodon.online]

@Da_Gut @zzt
... Sure, in the sense that you picked a file, but that's like saying you control what's in SPAM because you picked THIS can and not THAT one.

[zzt@mas.to]

@Da_Gut this is such a common thing that one of the multibillion dollar AI companies posted their official policy on it: using crafted prompts to extract the training data from an existing model of theirs is IP theft. of IP that was stolen from every web page ever.

that’s one of the mechanisms that will be used to make local LLMs a boon for the AI industrial complex rather than a threat. these supposed alternatives are already highly dependent on the system they claim to oppose.

[wackjackle@norden.social]

@MissConstrue @zzt 😭

[da_gut@dice.camp]

@zzt oh, I did not know that. So if you disconnect it from the Internet, it ceases to work?

[spinnyspinlock@infosec.exchange]

@zzt I love (hate) the "open source" LLMs lie

[zzt@mas.to]

@Da_Gut the factors I’m talking about aren’t technical ones, they’re social and systemic. specifically:

- local LLMs are worse than cloud ones, and necessarily must always be. it isn’t possible for independent development of models to happen, and LLMs are already on an intentionally fast deprecation cycle. old models aren’t viewed as useful by anybody.
- it’s very easy for established companies to take action against local models as IP theft, and they’re already laid the groundwork for this

[zzt@mas.to]

@Da_Gut - the hardware necessary to run local LLMs is being bought up en masse, in advance of its production, by AI companies, making local LLMs an avenue available only to the wealthy
- in any case, no such thing as an open source LLM exists. they are all binary blobs derived from proprietary and plagiarized data.

[bot_anix@nrw.social]

@MissConstrue @zzt das ist aber sehr ärgerlich!!!!