Thorjak
Brendan Nyhancc: everyone who repeated this
Ω 🌍 Gus Posey@brendannyhan Another way we'll destroy ourselves with AI nonsense, as we scramble to be the first to report it.
Tindra@brendannyhan @kevincollier + alt text:
Graham W. Jenkins Retweeted
Georgina Lee @lee_georgina
That story about the Al drone 'killing' its imaginary human operator? The original source being quoted says he 'misspoke' and it was a hypothetical thought experiment never actually run by the US Air Force, according to the Royal Aeronautical Society.
aerosociety.com/news/highlight...
Could an Al-enabled UCAV turn on its creators to accomplish its mission? (USAF
(UPDATE 2/6/23 - in communication with AEROSPACE - Col Hamilton admits he 'mis-spoke" in his presentation at the FCAS Summit and the rogue Al drone simulation' was a hypothetical
"thought experiment" from outside the military, based on plausible scenarios and likely outcomes rather than an actual USAF real-world simulation saying: "We've never run that experiment, nor would we need to in order to realize that this is a plausible outcome. He clarifies that the USAF has not tested any weaponised Al in this way (real or simulated) and says "Despite this being a hypothetical example, this illustrates the real-world challenges posed by Al-powered capability and is why the Air Force is committed to the ethical development of AI)
TC Won't Give In To Lies@brendannyhan
"We’ve never run that experiment, nor would we need to in order to realise that this is a plausible outcome,” Hamilton said.
...a plausible outcome 🫢
Lesley Carhart 
@brendannyhan @kevincollier everyone who was actually in the Air Force was like 🙃🤔
Kara Danvers 🇺🇦 
@brendannyhan i heard about this yesterday on a youtube video dont recall which one, something about the Ai targeting loctions and needs human to ok the target and the human said no and the AI wad mad or something and they went in re-wrote the code to obey the human and once online the ai took out the tower?
ᴚ uɐᗡ@brendannyhan wow
I found the article problematic as it wasn't clear until about halfway through that the claim wasn't even someone actually being hurt, but this makes it even worse.
Carrie@dannotdaniel @brendannyhan ok, I’m glad it wasn’t just me. I didn’t share for that reason. It seemed muddled and just “off.”
Nick 'The Viking'Oh, well, whew, I feel so much better now that I know that AI story was a mistake. I think I'll go pet that AI dog-robot I was wrong I've misjudged it's cuteness.
THIS ACCOUNT HAS MOVED
Dave Alvarado@brendannyhan "mis-spoke" meaning "leaked classified intel", probably.
Bee O'Problem 
@brendannyhan oddly I missed out on The Discourse on this by seeing the headline and assuming a sequence like
* Dude summons drone
* Drone returns
* Misuse or malfunction caused drone to fatally crash into dude instead of landing properly
Serves me right for assuming news was about stuff that's exists somewhere in the vicinity of plausible
Square Four FoursMicro SF/F by O. Westin (@[email protected])
"We made a simulation," the colonel said. "Did you?" the robot said. "Well, a thought experiment." "And what did you think?" "That you could, in theory, kill someone." The robot looked at its bonds. "I was thinking the same thing." "That you could kill someone?" "That you could." #MicroFiction #SmallStories #TootFic
Old One Eye@brendannyhan Frankly hasn't this scenario existed in theoretical literature for decades?
The more shocking part of the exercise described would have been if they hadn't run simulations prior to what was described to specifically try to identify ways the AI could go "off-script" and cause harm.
Probably Paul@brendannyhan Thanks for this. I've un-shared the story from my TL.
Starry Starry Knight
Horror Short Film “Slaughterbots” | ALTER
@Space6host
@brendannyhan You may also enjoy this excellent episode of Black Mirror, which may still be on Netflix.
https://youtu.be/BDP9jA6k_UE
@brendannyhan You may also enjoy this excellent episode of Black Mirror, which may still be on Netflix.
https://youtu.be/BDP9jA6k_UE
Black Mirror (2017) Metalhead. Deadliest Dogs. Robot Dogs.
Austin PhilpHonestly it seemed really fishy from the get-go given that it was almost a textbook example of a value-loading problem.
It's the kind of thing you see in cheap SciFi or philosphical thought experiments, it's certainly not the sort of thing that an actual engineer working with AI/ML would overlook in practice
Wendy Lady@brendannyhan I read the article and reread it to make sure I understood it was a hypothetical, not literal circumstance. That being said, it was poorly articulated in the article unless you paid close attention.
catastrophile@brendannyhan somebody overheard a movie pitch and mistook it for actual events
WndlB@brendannyhan otoh, send thoughts and prayers to the funeral pyre of that Colonel's career...
Phra Nuntiyo@brendan yes, i expected him to be crypto-fascst warmonger but not nearly to this extent!
A Tattered Scrapbook@brendannyhan Trillions in military spending…
Patricia Cordeiro@brendannyhan I read this yesterday, and I thought it could be false because it reminded me a bit of this short animation:
DRONE | Animated short film about drones, AI, and live-streaming on YouTube
Dan Gillmor@brendannyhan It was amazingly fishy from the start...
fung0@brendannyhan Sorry - which version of this story am I meant to disbelieve?