this post was submitted on 04 May 2026
106 points (94.9% liked)

Technology

84324 readers
6407 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 news or articles.
  3. Be excellent to each other!
  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, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
  9. Check for duplicates before posting, duplicates may be removed
  10. Accounts 7 days and younger will have their posts automatically removed.

Approved Bots


founded 2 years ago
MODERATORS
top 27 comments
sorted by: hot top controversial new old
[–] GreenBeanMachine@lemmy.world 58 points 7 hours ago (2 children)

38.6% of respondents said they "very trust" or "somewhat trust" advice from AI-generated sources regarding relationships and social interactions.

Very worrying.

[–] HalfSalesman@lemmy.world 2 points 46 minutes ago

TBH, I'm an autistic male and using AI for dating advice has 100% helped me get laid (still working on a stable partner though). Not because its super smart, but because I can rapidly ask multiple questions I'd be otherwise unable to ask a human without incredible and crippling embarrassment and get answers that are "good enough" if a tad broad or obvious probably to most people.

But as a broad trend, I'll admit that is a little worrying.

[–] TubularTittyFrog@lemmy.world -1 points 4 hours ago* (last edited 4 hours ago) (1 children)

AI responses are probably less shit than they'd get from real people these days.

half the people I meet now are living in a delusion-fueled alternate reality and suffering from main character syndrome, they are utterly useless for any social or relationship advice.

and social media is 90% terrible advice pushed my influencers who are only there to stir up drama or 'work on yourself', both of which are pretty useless advice, and are just designed to get you hooked on buying shitty products they advertise.

[–] GreenBeanMachine@lemmy.world 17 points 4 hours ago* (last edited 4 hours ago) (1 children)

and social media is 90% terrible advice

Well, I hope and pray that no one ever trains AI on social media posts and then they regurgitate the same bullshit.

Oh...wait...

half the people I meet now are living in a delusion-fueled alternate reality and suffering from main character syndrome

I'm very sorry you're surrounded by such terrible people, but I disagree with you. People understand people better than machines.

[–] magnue@lemmy.world 23 points 8 hours ago (3 children)

I only consult AI about my homelab problems.

[–] egrets@lemmy.world 2 points 41 minutes ago

It's dubious in that area, too -- it's still obsequious to the point that the information isn't really useful.

"Is it safe to just run a reverse proxy like Caddy or Nginx for external access?" -> "Yes, that's safe and secure."

"Shouldn't I also use a separate VLAN on my home network?" -> "Great point! It's much safer to use a dedicated VLAN."

"Are you sure I shouldn't set up a VPN to tunnel into my home network?" -> "Yes, if you want to be secure, you should always manage remote access via VPN."

[–] ramble81@lemmy.zip 15 points 5 hours ago* (last edited 5 hours ago)

Sadly that’s a big problem. “Where did it get its data from?”…”Stack Exchange?”…”Cool, you gonna post new answers there or that it worked?”…”No”…”So where does it get its data from for the next thing?”

It becomes an oroborous loop that causes less and less information to be available. You can’t exactly feed it a manual and expect it to understand. Not to mention a lot of things don’t have manuals or good documentation. It’s like Discord, but worse.

[–] mynameisbob@lemmy.ml 14 points 7 hours ago (1 children)

and most of the time it doesn't help.

[–] magnue@lemmy.world 0 points 2 hours ago

It literally set up about 20-25 services in a week and I'm not experienced with it.

[–] thirdBreakfast@lemmy.world 4 points 5 hours ago
[–] mynameisbob@lemmy.ml 2 points 7 hours ago
[–] BarbecueCowboy@lemmy.dbzer0.com 1 points 7 hours ago* (last edited 7 hours ago) (1 children)

Around non-specifically 50% compared to as high as 29.999% doesn't seem like a huge delta. I wonder what the real number on both is.

Edit: Looks like it was 52.4 and 29.1 respectively. The numbers on how much they trust the responses later in the article might be more worrying.

[–] Buffalox@lemmy.world 2 points 6 hours ago* (last edited 6 hours ago) (2 children)

That's because you don't understand the numbers. 30% to 50% means it's actually 66% more common for girls.

[–] Barbarossa@piefed.ca 4 points 4 hours ago (1 children)

And with the actual numbers of 52.4% and 29%, its closer to 80% increase in how much women trust the responses. That’s a huge delta, what is driving that

[–] wols@lemmy.zip 4 points 3 hours ago

Random uninformed guess: women are on the whole more open to sharing their problems to begin with. So it's probably less "teenage girls turn to AI with their problems while men go to therapy" and more "men continue to avoid talking about their problems to anyone, including LLMs".

It is worrying that social interactions and support are getting delegated to algorithms at such high rates but I'm not convinced there is a significant gender gap to be explained on the technology level. Dudes probably ask the word prediction machine plenty about cars, tech, or weird conspiracy theories.

[–] Jax@sh.itjust.works 0 points 5 hours ago (3 children)
[–] atomicbocks@sh.itjust.works 2 points 3 hours ago (1 children)

This person appears to be using shrinkflation math: The difference between 30 and 50 is 20, 20 is 66% of 30. When you see a bottle on the counter in the store that says 20% more that’s how they’re getting their number. It’s the percentage of the smaller bottle.

[–] Jax@sh.itjust.works 1 points 1 hour ago

The math was not the funny part, the funny part was them discussing something that had already been included in the article.

When asked about their purposes for using AI-generated content in daily life (multiple answers allowed), the most common overall response was information retrieval and research at 76.4%, followed by writing and editing texts at 33.9%, and seeking advice at 23.3%. Information retrieval remained the top choice across all age groups, but seeking advice ranked second among teenage girls and third among women in their 20s to 40s and men in their 30s.

38.6% of respondents said they "very trust" or "somewhat trust" advice from AI-generated sources regarding relationships and social interactions. Trust levels were higher among younger generations, with over half of those in their teens and twenties expressing this trust. Among teenage girls, the figure reached 63.1%.

As I say above, it's funny to me that neither of them translated the article.

[–] Buffalox@lemmy.world 0 points 2 hours ago (1 children)

No absolutely not. This is reality, if something increases from 30% to 50%, you need to increase your capacity to handle it by 66%.
That's reality, and not moronic idiocy.

[–] Jax@sh.itjust.works 1 points 2 hours ago* (last edited 2 hours ago) (1 children)

Correct, my question is because the article outright states the number:

38.6% of respondents said they "very trust" or "somewhat trust" advice from AI-generated sources regarding relationships and social interactions. Trust levels were higher among younger generations, with over half of those in their teens and twenties expressing this trust. Among teenage girls, the figure reached 63.1%.

Going off saying 'you don't understand the numbers' when neither of you have translated the article seemed genuinely funny to me.

Edit: I should say, I recognize you understand the numbers — I was not calling out your math.

[–] Buffalox@lemmy.world 0 points 1 hour ago* (last edited 1 hour ago)

I did translate and read the article, so WTF do you mean?

I used rounded numbers that are close to make the calculation easier to follow for the people that apparently don't understand how this works.

The part you quote was already quoted.

[–] Aquifel@sh.itjust.works -2 points 5 hours ago* (last edited 4 hours ago) (2 children)

Dude probably reversed some numbers, math is hard sometimes. Or... they're focusing only on comparing between the affected population which is kinda weird.

[–] Jax@sh.itjust.works 1 points 2 hours ago

No, their math is right — it says so in the article itself.

[–] wols@lemmy.zip 5 points 3 hours ago* (last edited 3 hours ago) (1 children)

Well the math they did was 0.5/0.3 = 1.(6)

To make the logic for that math easier to follow, imagine it was actually 60% of teenage girls rather than the 50% from the article.
If you pick a random man, there is a 30% chance they consult AI. If you pick a random girl, that chance is instead 60%. So twice as likely, or expressed a different way, 100% more likely than when picking a random man.

Switching back to the 30/50 numbers you get that a random teenage girl is (at least) 66% more likely to turn to AI than a random man.
To me, this seems like a reasonable way to compare these numbers and it makes it clear that the difference is actually pretty significant, contrary to OP comment's claim.

[–] Buffalox@lemmy.world 0 points 1 hour ago

Absolutely 100% correct, and if I recall correctly, it's about 5th grade math. It's astounding the number of people here who don't understand such a simple concept.