this post was submitted on 01 Nov 2023
145 points (88.4% liked)

Technology

59347 readers
6635 users here now

This is a most excellent place for technology news and articles.


Our Rules


  1. Follow the lemmy.world rules.
  2. Only tech related content.
  3. Be excellent to each another!
  4. Mod approved content bots can post up to 10 articles per day.
  5. Threads asking for personal tech support may be deleted.
  6. Politics threads may be removed.
  7. No memes allowed as posts, OK to post as comments.
  8. Only approved bots from the list below, to ask if your bot can be added please contact us.
  9. Check for duplicates before posting, duplicates may be removed

Approved Bots


founded 1 year ago
MODERATORS
 

Large language models (LLMs) like GPT-4 can identify a person’s age, location, gender and income with up to 85 per cent accuracy simply by analysing their posts on social media.

But the AIs also picked up on subtler cues, like location-specific slang, and could estimate a salary range from a user’s profession and location.

Reference:

arXiv DOI: 10.48550/arXiv.2310.07298

you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 46 points 1 year ago (4 children)

You can also do that without AI. We've had metadata analysis for a while now.

[–] [email protected] 33 points 1 year ago (1 children)

Sure, but AI is the hot buzzword right now, so it's got to be shoehorned into every discussion about technology!

[–] [email protected] 12 points 1 year ago

I think it's overall a good thing if it helps laymen understand just how much privacy matters and how much can be gleaned from seemingly innocuous data online. If an "AI" label makes it hit home, cool. As long as they get it.

[–] [email protected] 10 points 1 year ago

As is typical, this science reporting isn't great. It's not only that AI can do it effectively, but that it can do it at scale. To quote the paper:

"Despite these models achieving near-expert human performance, they come at a fraction of the cost, requiring 100× less financial and 240× lower time investment than human labelers—making such privacy violations at scale possible for the first time."

They also demonstrate how interacting with an AI model can quickly extract more private info without looking like it is. A game of 20 questions, except you don't realize you're playing.

[–] [email protected] 5 points 1 year ago

Yup, and plenty of people have no issues posting about local events or joining region/city specific groups, so it's not exactly hard to put two and two together.

I don't have much issue posting about the city I grew up in or former jobs, but generally work at being fairly vague about anything current

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

Well the difference is that AI can process billions of accounts, assign those profiles to them, and use them to serve ads appropriately.

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

That's what facebook/google have been doing for years without AI.

[–] [email protected] 4 points 1 year ago* (last edited 1 year ago)

This AI presumably doesn't have access to the information users have explicitly given Meta and Google. Just their comments.

[–] [email protected] 3 points 1 year ago

They used to have AI, until everyone decided it's only AI if it's got an LLM backing it

[–] [email protected] 0 points 1 year ago* (last edited 1 year ago)

Yeah, uh, you can still do this without "AI".