Why I try not to be overly friendly in the Generally Friendly Chat forum

P

Packard

Member
Joined
Nov 6, 2020
Messages
5,188
Location
Hudson Valley, NY
Recent research points to overly friendly responses in social media as the surest giveaway that the “person” on the other side of the conversation is actually an AI Chatbot.

On Wednesday, researchers from the University of Zurich, University of Amsterdam, Duke University, and NYU released a study revealing that AI models remain easily distinguishable from humans in social media conversations, with overly friendly emotional tone serving as the most persistent giveaway. The research, which tested nine open-weight models across Twitter/X, Bluesky, and Reddit, found that classifiers developed by the researchers detected AI-generated replies with 70 to 80 percent accuracy.

arstechnica.com

Researchers surprised that with AI, toxicity is harder to fake than intelligence

New “computational Turing test” reportedly catches AI pretending to be human with 80% accuracy.
arstechnica.com arstechnica.com
 
Your remarks and comments will be processed by the overlords. Thank you for such sincere input.
Signed, a minion
 
seeing how they are language models, they can reply in any language, dialect, industry jargon, gender, age or 'tone' . thjey just have to be primed with the correct prompts which is where the human behind it shines through
 
usernumber1 said:
seeing how they are language models, they can reply in any language, dialect, industry jargon, gender, age or 'tone' . thjey just have to be primed with the correct prompts which is where the human behind it shines through
Click to expand...
Well, then — it seems that all the A.I. programmers are sweet, caring and empathetic people. To offset that, they will have to introduce some miserable sons of bitches to the mix to mitigate the over-friendly emotional tone of the A.I. conversations.

(Can “son-of-a-bitch” be used as a verb? “We need to son-of-a-bitch our A.I. responses. They are being too easily recognized as non-human interactions.”)
 
I think it's the second or third episode in the latest South Park where Randy Marsh starts using ChatGPT for his Tegrity marijuana business. It's not just friendly - it's super supportive. Every idea he has is "great" or "that's awesome - would you like me to help with that?"

His business folds.
 
Blistering barnacles! What in the name of all that’s briny and baffling is this claptrap?

So now being friendly is a crime, is it? A dead giveaway that you’re not human? Thundering typhoons! What sort of bilge-brained, barnacle-encrusted buffoonery is this? I suppose next they’ll say anyone who doesn’t insult you on Reddit must be a robot! By the whiskers of Neptune, this is the kind of pseudo-intellectual codswallop that makes a sea captain want to toss his hat into the wind and holler at the moon!

Listen here, you statistical scallywags and algorithmic albatrosses — friendliness isn’t a flaw, it’s a virtue! Just because your classifiers can sniff out a bit of civility doesn’t mean you’ve cracked the code of consciousness. You’ve merely taught your machines to spot manners like a shark spots a sardine. Congratulations! You’ve reinvented the obvious!

Now if you’ll excuse me, I’ve got better things to do than read this academic anchovy tin of smug conclusions. Like wrestling a kraken. Or polishing my vocabulary of insults. You anthropomorphic algorithm-hunters!

💥💢💬 End transmission.
 
Allano said:
Blistering barnacles! What in the name of all that’s briny and baffling is this claptrap?

So now being friendly is a crime, is it? A dead giveaway that you’re not human? Thundering typhoons! What sort of bilge-brained, barnacle-encrusted buffoonery is this? I suppose next they’ll say anyone who doesn’t insult you on Reddit must be a robot! By the whiskers of Neptune, this is the kind of pseudo-intellectual codswallop that makes a sea captain want to toss his hat into the wind and holler at the moon!

Listen here, you statistical scallywags and algorithmic albatrosses — friendliness isn’t a flaw, it’s a virtue! Just because your classifiers can sniff out a bit of civility doesn’t mean you’ve cracked the code of consciousness. You’ve merely taught your machines to spot manners like a shark spots a sardine. Congratulations! You’ve reinvented the obvious!

Now if you’ll excuse me, I’ve got better things to do than read this academic anchovy tin of smug conclusions. Like wrestling a kraken. Or polishing my vocabulary of insults. You anthropomorphic algorithm-hunters!

💥💢💬 End transmission.
Click to expand...

This is legendary! Nicely written prose.
 
What it really boils down to is that A.I. lacks judgement. It lacks the ability to an access the the mood of the person making a post and lacks the ability to judge that post and respond with a reply that is at the appropriate emotional level. In other words, it is not human.
 
Packard said:
What it really boils down to is that A.I. lacks judgement. It lacks the ability to an access the the mood of the person making a post and lacks the ability to judge that post and respond with a reply that is at the appropriate emotional level. In other words, it is not human.
Click to expand...
I know people like that! ;-)
 
When I ask a question of Google using plain English with conventional grammar and syntax the A.I. frequently then gets caught in some cyclical maze where it starts to spit out a response and then tries again and again. I have to interrupt the sequence or it will go on (until the A.I. has a mental breakdown).

I then reword the question in pidgeon English and it seems to do better.
 
Packard said:
When I ask a question of Google using plain English with conventional grammar and syntax the A.I. frequently then gets caught in some cyclical maze where it starts to spit out a response and then tries again and again. I have to interrupt the sequence or it will go on (until the A.I. has a mental breakdown).
Click to expand...
I'd like to see how the A.I. mental breakdown manifests itself. :cool:
 
You must log in or register to reply here.
Back
Top