“AI-powered writing tools are increasingly integrated into our e-mails and phones.
-
"We wanted to rigorously test whether risky behaviors, such as confirming delusional beliefs, assuming clinical authority, or gradually eroding boundaries, can emerge through multi-turn interactions," so Youyou Cheng, first author.
"By demonstrating that such failures do occur and can be systematically elicited, the paper establishes the need for structured safeguards…"
@rexi
Sam Altman: Why are you showing me this? My housekeeper already stocked up on toilet paper last week!
@gregeganSF#aiethics meets #capitalism
-
“AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs”
“We told people before, and after, to be careful, that the AI is going to be (or was) biased, and nothing helped,” Naaman said. “Their attitudes about the issues still shifted.”
AI autocomplete doesn’t just change how you write. It changes how you think
AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs
Scientific American (www.scientificamerican.com)
@gregeganSF This is what I have always been concerned with and is why I have never used any auto-complete, well before LLMs arrived.
It doesn't even need to be insidious, we instinctively pick up cues. Every time Microsoft re-enables auto-complete on my Outlook account, I can feel my writing change until I diable it. Maybe my writing could change for the better, but if I do not like how AI writes, why would I want to be influenced by it?
That said I also do not trust these companies at all.
-
“AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs”
“We told people before, and after, to be careful, that the AI is going to be (or was) biased, and nothing helped,” Naaman said. “Their attitudes about the issues still shifted.”
AI autocomplete doesn’t just change how you write. It changes how you think
AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs
Scientific American (www.scientificamerican.com)
@gregeganSF Eek, this is scary
-
“AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs”
“We told people before, and after, to be careful, that the AI is going to be (or was) biased, and nothing helped,” Naaman said. “Their attitudes about the issues still shifted.”
AI autocomplete doesn’t just change how you write. It changes how you think
AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs
Scientific American (www.scientificamerican.com)
@gregeganSF this is in line with research showing that implicit bias training doesn't actually reduce bias
edit: I'm allergic to not providing sources and broke out in hives until adding these links
-https://www.scientificamerican.com/article/the-problem-with-implicit-bias-training/
- https://www.science.org/doi/10.1126/sciadv.ado5957
- https://www.ncbi.nlm.nih.gov/books/NBK603840/ -
“AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs”
“We told people before, and after, to be careful, that the AI is going to be (or was) biased, and nothing helped,” Naaman said. “Their attitudes about the issues still shifted.”
AI autocomplete doesn’t just change how you write. It changes how you think
AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs
Scientific American (www.scientificamerican.com)
@gregeganSF more information in the article Scientific American based their short article on:
https://www.eurekalert.org/news-releases/1119239 -
“AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs”
“We told people before, and after, to be careful, that the AI is going to be (or was) biased, and nothing helped,” Naaman said. “Their attitudes about the issues still shifted.”
AI autocomplete doesn’t just change how you write. It changes how you think
AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs
Scientific American (www.scientificamerican.com)
@gregeganSF Just wait for the day you can’t turn this off and the day where you are not allowed to write on your own anymore. Just to please those fucking TechBros #ButilerianJihadNow
-
@mrgtwentythree @gregeganSF If you want to just lie down in the shit and die, go ahead. The rest of us don't see passively accepting fascist conditioning as okay, though.
@ivy_pigeon95 @gregeganSF waffles or pancakes
-
@ivy_pigeon95 @gregeganSF waffles or pancakes
@mrgtwentythree @gregeganSF Right, you're a dumbass troll. Got it.
-
“AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs”
“We told people before, and after, to be careful, that the AI is going to be (or was) biased, and nothing helped,” Naaman said. “Their attitudes about the issues still shifted.”
AI autocomplete doesn’t just change how you write. It changes how you think
AI-powered writing tools are increasingly integrated into our e-mails and phones. Now a new study finds biased AI suggestions can sway users’ beliefs
Scientific American (www.scientificamerican.com)
@gregeganSF
Ugh. I guess it follows, but is not good news. I've been (maybe naively) continuously surprised how folks will allow things like spelling and grammar type checkers (I'm looking at you, Word) decide on wording or spelling choices that are, imo, really stylistic or voice decisions -
@gregeganSF These are the Thinking Machines from Dune from 60 years ago.
Not killer robots that hunt down humans to enslave and murder them. But billionaires conditioning lazy and gullible people to make them obedient to their will.@yora @gregeganSF try to take a slightly more generous view. People aren't lazy and gullible; they're completely stressed, overworked, bamboozled and alienated. This is fertile ground for technofascism.
-
@yora @gregeganSF try to take a slightly more generous view. People aren't lazy and gullible; they're completely stressed, overworked, bamboozled and alienated. This is fertile ground for technofascism.
@fluidlogic @gregeganSF None of that are reasons to use AI.
-
@fluidlogic @gregeganSF None of that are reasons to use AI.
@yora @gregeganSF exactly right.
-
E em0nm4stodon@infosec.exchange shared this topic
