this post was submitted on 24 Jul 2024
0 points (NaN% liked)

Memes

45731 readers
71 users here now

Rules:

  1. Be civil and nice.
  2. Try not to excessively repost, as a rule of thumb, wait at least 2 months to do it if you have to.

founded 5 years ago
MODERATORS
gary_host_laptop@lemmy.ml
sexy_peach@feddit.de
cyclohexane@lemmy.ml
cypherpunks@lemmy.ml
0
Yet another good recipe (lemmy.ml)
submitted 4 months ago by xelar@lemmy.ml to c/memes@lemmy.ml
54 comments fedilink hide all child comments
 
you are viewing a single comment's thread
view the rest of the comments
[–] JusticeForPorygon@lemmy.world 0 points 4 months ago (4 children)

So I'm assuming these boys are built through ChatGPT, so, if that's the case, how are they able to say things like this? I'm 99% sure if I asked ChatGPT to defend the Russian invasion of Ukraine it would tell me to fuck off.

Is there another chatbot that is capable of being a bigot?

[–] ingeniosissimo@lemmy.world 0 points 4 months ago

Plenty of text generation models can be run on consumer desktop/gaming hardware with fairly acceptable output. There's probably also several models fine-tuned with uncensored data, that will spew any kind of propaganda you ask of it. Hugging Face hosts many publicly available models.

[–] the_wise_wolf@feddit.org 0 points 4 months ago

There are a few: https://en.m.wikipedia.org/wiki/List_of_large_language_models

You can run some of them on your computer and do anything with them.

Others - Grok in particular - don't have safeguards. And even if they have safeguards, depending on how they were implemented, you may be able to circumvent them.

[–] ogeist@lemmy.world 0 points 4 months ago

There are open LLMs models, you can have your own very biased assistant.

[–] breakingcups@lemmy.world 0 points 4 months ago

Despite what others have mentioned, running a different LLM locally, it's also possible to get ChatGPT to do this sort of stuff by telling it to participate in a "debate exercise" and giving it its talking points.