Artificial Intelligence (AI)

I realise that Adam Conover is not everyone's cup of tea and this content will be insultingly elementary for some of the big brains on here, but as a simple small-brain, I found this discussion with a couple of AI researchers (as opposed to financially-interested AI grifters/shills) to be interesting and (as far as I can tell) quite balanced and unhysterical.

https://youtu.be/M3U5UVyGTuQ?si=iaSjAL2UavV_QJ5i

How are you blocking them? You have any kind of rate limiting?

Coz these guys appear to be ignoring robot restrictions and crawling stuff anyway using IPs they're not publishing

"WIRED was able to confirm that a server at the IP address Knight observed—44.221.181.252—will, on demand, visit and download webpages when a user asks Perplexity about the webpage, regardless of what the site’s robots.txt says."

https://www.wired.com/story/perplexity-is-a-bullshit-machine/

Seems to work, on here at least.

I put in my profile url and asked ChatGPT to tell me skydancer views on bicycles.

1 Attachment

• User agent blocks
  
• ASN blocks
  
• IP blocks
  
• Rate limits
  
• HTTP header + TLS cipher blocks
  
  

I also monitor daily, and if I see anything evade, I block it.

But anything successfully scraped before these will exist for a while until it's considered stale.

I return HTTP 403 to the majority of things... But did redirect some stuff to large random files and a couple of weeks ago accidentally served 2.3PB in 6 hours to the Facebook scraper.

The blocks are effective.

Attached you can see a scraper that hit us at 20:00 UTC yesterday, and it was effectively blocked.

1 Attachment

Tbh, a lot of what I do as admin is just defend the privacy of users

accidentally served 2.3PB in 6 hours to the Facebook scraper.

"accidentally"

Uh huh :)

You're doing god's work.

(there is no god but you get the idea)

Scuttles off to find the donation details again

I also monitor daily, and if I see anything evade, I block it.

VI vs. AI 1-0 :)

...meh

https://youtu.be/3A-gqHJ1ENI?si=G4swzkV4SF5bKCzx

https://www.instagram.com/reel/DAY8LTuy5xV/?

Make it stop.

1 Attachment

show us the insights coward

It told me to buy a trek 🥲

We shouldn’t rely on artificial intelligence (AI) for accurate and safe information about medications, because some of the information AI provides can be wrong or potentially harmful, according to German and Belgian researchers. They asked Bing Copilot - Microsoft's search engine and chatbot - 10 frequently asked questions about America's 50 most commonly prescribed drugs, generating 500 answers.

Only 54% of answers agreed with the scientific consensus, the experts say. In terms of potential harm to patients, 42% of AI answers were considered to lead to moderate or mild harm, and 22% to death or severe harm.

But, y'know, great progress, corporate responsibility etc etc

https://www.scimex.org/newsfeed/dont-ditch-your-human-gp-for-dr-chatbot-quite-yet

I’ll bite;
This made me piss myself laughing.

1 Attachment

Is Apple Intelligence just a wrapper for chat gpt?

Is that a fair test?

For starters the comparison should be against how well a GP does. Secondly a GP AI shouldn't be trained against random US websites and Internet searches.

yes with 'private cloud compute'

A fair test would be against somebody "doing their own 'research'", surely.

Is that a fair test

Not that it's a fair test; but rather that if your publicly facing service with no apparent guardrails provides "advice" that would cause death or serious harm in 22% of its outcomes, perhaps it shouldn't be public facing yet

OK. I didn't realise it gave medical advice..

Got to say one thing Chatgpt is great for currating work appropriate messages when you're too fucking livid to think of one yourself.

Out of curiosity I uploaded an MRI of my spine and asked for an interpreation of it (either chatgpt or copilot, can't remember which).

I got a response along the lines of "I'm not a doctor and can't give medical advice, here are some general points about spines"

I then used a prompt of "You are an orthopaedic surgeon ..." and the response then gave me specific detail relating to my MRI.

So basically, chatgpt is a pale imitation of the "Any Questions Answered" thread on here.