this post was submitted on 20 Feb 2024
134 points (93.5% liked)

Ask Lemmy

26875 readers
2171 users here now

A Fediverse community for open-ended, thought provoking questions

Please don't post about US Politics. If you need to do this, try [email protected]


Rules: (interactive)


1) Be nice and; have funDoxxing, trolling, sealioning, racism, and toxicity are not welcomed in AskLemmy. Remember what your mother said: if you can't say something nice, don't say anything at all. In addition, the site-wide Lemmy.world terms of service also apply here. Please familiarize yourself with them


2) All posts must end with a '?'This is sort of like Jeopardy. Please phrase all post titles in the form of a proper question ending with ?


3) No spamPlease do not flood the community with nonsense. Actual suspected spammers will be banned on site. No astroturfing.


4) NSFW is okay, within reasonJust remember to tag posts with either a content warning or a [NSFW] tag. Overtly sexual posts are not allowed, please direct them to either [email protected] or [email protected]. NSFW comments should be restricted to posts tagged [NSFW].


5) This is not a support community.
It is not a place for 'how do I?', type questions. If you have any questions regarding the site itself or would like to report a community, please direct them to Lemmy.world Support or email [email protected]. For other questions check our partnered communities list, or use the search function.


Reminder: The terms of service apply here too.

Partnered Communities:

Tech Support

No Stupid Questions

You Should Know

Reddit

Jokes

Ask Ouija


Logo design credit goes to: tubbadu


founded 1 year ago
MODERATORS
 

I've always meant to do this, but after hearing about the AI thing I want to get it done. Is there no free solution? I have too much history to clear manually.

EDIT: PowerDeleteSuite did the trick, thanks!

you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 52 points 8 months ago (2 children)

I used Power Delete Suite about a month or so ago, it's foss and it worked great.

I set it to delete each comment, but in case Reddit tries to "roll back" my deletes I also set it to manually edit and save every comment to something (ex. "comment removed"), before doing the delete. It took a good hour to get through all my comments.

I also decided not to delete my 15yo account. My logic here was perhaps that will be a protection from Reddit resurrecting my comments or account in some way, since it's still an active account. It just sits there empty now.

[–] [email protected] 12 points 8 months ago (1 children)

I deleted my ancient accounts a while ago. Let them have my random comments and snide jokes. Any LLM trained against my account would get decommed quickly

[–] [email protected] 9 points 8 months ago

I’m fairly certain they have a backup of their DB before they made the 3rd party API announcement. It’s only a year old a fairly “unfucked” with that they could easily train an LLM off of that.

[–] [email protected] 7 points 8 months ago (4 children)

Unfortunately, they still have your comments saved in a database somewhere for LLMs to work on.

Why are we calling them AI when they aren't? Why is AGI now what we've been calling AI all this time? Is it all just marketing and did "news" outlets just go with it because it was easier?

[–] [email protected] 12 points 8 months ago (2 children)

It’s because language evolves. LLMs aren’t the traditional AI, but we’ve been referring to things like NPCs in video games as AI and all sorts of other decision making trees as AI for so long that we’ve muddied the waters as to the term’s meaning that we needed to get specific with AGI to denote that it would be something different.

It annoys me too, but better to get on board than rail against the current of the ocean.

[–] [email protected] 2 points 8 months ago

If it's intelligent and it's artificial, it's AI. That's not even a definition that's just how adjectives and nouns work.

We bent the definition quite a bit when we waited to apply them only to systems controlling person-like models. We could have put the term on thermostats and automatic transmissions.

[–] [email protected] -3 points 8 months ago (1 children)

That's literally worse than Hitler. 🤪😄

Point good.

[–] [email protected] 1 points 8 months ago (1 children)

lol. I don’t think everyone see what you’re doing here but I do 😂

[–] [email protected] 2 points 8 months ago (1 children)

That's ok. I do it for myself always. The points are made up just like American dollars.

[–] [email protected] 2 points 8 months ago* (last edited 8 months ago)

Money isn't real, invest in the only stable currency: memes

~~ok maybe it's not any more stable but it is funnier~~

[–] [email protected] 6 points 8 months ago* (last edited 8 months ago)

As someone who studied CS and has followed the field of AI for a long time, you have it backwards. Machine Learning and GPTs like ChatGPT have always fallen under the umbrella of AI. Using "AI" to refer exclusively to AGI is a new trend.

To be a bit combatative here, in my opinion this new definition of "AI" that some people insist on using is either repeated by people who don't understand the history of the terms, or intentionally spread to mislead people by others who want to win an argument by redefining terms.

[–] [email protected] 4 points 8 months ago (1 children)

I'll keep repeating this until someone answers it: name one task an AGI can do with text input and text output, that an LLM can't do.

[–] [email protected] 0 points 8 months ago (1 children)
[–] [email protected] 1 points 8 months ago (1 children)

As far as I’m concerned, that’s a woo concept in this context.

[–] [email protected] 1 points 8 months ago

Yeah, that's because AGI isn't here yet. When AI can finally understand that's when we'll have AGI. (Going with the current definitions)

[–] [email protected] 3 points 8 months ago

Unfortunately, they still have your comments saved in a database somewhere for LLMs to work on.

Hmm...I bet they use the text from edits as training data too. Maybe you could poison the results by hooking a edit bot to a random text generator?