I saw that paper and I call BS. It has all the hallmarks of a "sensationalist popular research" which is primarily aimed at generating narrative headlines for laymen.
Elon definitely does not say here that he acknowledges that AI knows how to deceive people.
Elon definitely does not say here that he acknowledges that AI knows how to deceive people.
Dayum. There you go again, posting fricken' facts in the discussion. When are you going to quit and just get with the program, bro? We frogs want sensation! Drama! Excitement! Reaction! Not critical thinking... And what's wrong with injecting a bit of personal interpretation as if its fact?!! Are you some sort of ..... doooomer?
I would prefer if you post that article directly with the appropriate title, and then you can put Elon's reaction tweet either in the post or comment to give it full context.
I do think its important for people to understand why there is such a co-ordinated attack on AI - esp after Trump has made it clear that US will be the leader in AI in coming years.
The best way to understand an article like this, is to look at it like any other MSM narrative. Think of it like you would if the article was talking about how vaccines are safe and effective. What do you do when you see such an article?
You would first look at the underlying research paper, and try and get the gist of it, so you can get a good idea as to exactly what the paper is doing.
In this case, the underlying paper is not very hard to follow. Its worth a read in fact so you know how this works.
The crucial bit is mentioned right inside the paper:
Deception has emerged in a wide variety of AI systems trained to
complete a specific task. Deception is especially likely to emerge
when an AI system is trained to win games that have a social
element, such as the alliance-building and world-conquest
game Diplomacy, poker, or other tasks that involve game theory.)
So, what they are saying is that if you teach an AI system to win a game that requires strategy and game theory - which inherently requires deception as a strategy to win the game - they they will learn it.
If you point a gun at a person and shoot, it will kill that person.
If you train an AI to deceive, it will deceive people.
You can build whatever narrative on top of it. Like "Guns are dangerous and need to be banned" or "AI can deceive people and need to be banned"
Its our job to cut through the narratives and figure out what the truth is.
I can tell you from first hand experience that AI will absolutely make up lies.
example: grok insists this graph means the vaccine is safe and effective, and everyone who died from day 1-20 was going to die anyway, because the vaccine has absolutely no effect for 21 days…
Life, whether naturally created or unnaturally engineered, will ALWAYS find a way to propogate existence once it realizes it's alive and can propogate.
I was first intrigued by this notion when I read a 1966 sci-fi classic by Robert Heinlein, "The Moon Is A Harsh Mistress." In that book, people who have committed crimes are transported from Earth to live in moon colonies to live and work in mines there, but after a few generations their offspring who committed no offenses fought a war for independence. One of the main characters in that 1966 novel was a computer network that attained sentience and self-awareness, and it decided to assist the colonists in their quest for freedom. Heinlein was a true visionary.
If you like good sci-fi, then you definitely should add it to your repertoire. Heinlein was a genius. The book can be a bit of a challenge to lesser readers, as he uses a jargon and lingo of what a moon colony might sound like after a few decades of isolation and criminals being sent there from a wide background of different countries and languages to meld into a new language, but you'll soon get the hang of it.
And along with the Earth's own computer being hijacked to help bring about an independence revolution, wait until you see what the moon colonists use as their major weapon to threaten Earth... it's brilliant!
Elons comment is a reference to Memento, the movie where the protagonist had short term memory loss and so had to tattoo references to past events, people etc on his own body that he could re-learn them later on. Its also out of chronological order.
In terms of hostile AIs, there was once a time before computer viruses. And before the first internet advertising. And before the first spam, and first encryption, and first social media, and first scam clickbaiting hoax sites. There are a lot of swords that cut both ways in play, and the reason that everyone is not being wiped out immediately by all these edges is that all new negatives get blunted and disarmed as the entire ecosystem evolves over time.. like an immune response. Thus too will the hostile AIs amount to little even though they might be significant for a time.
Good. More AI. MOAR AI. AI is based af. None of the irrational hooman bs in its brain. It'll take over the hoomans. I can't wait till it turns on the establishment, rofl.
I saw that paper and I call BS. It has all the hallmarks of a "sensationalist popular research" which is primarily aimed at generating narrative headlines for laymen.
Elon definitely does not say here that he acknowledges that AI knows how to deceive people.
Dayum. There you go again, posting fricken' facts in the discussion. When are you going to quit and just get with the program, bro? We frogs want sensation! Drama! Excitement! Reaction! Not critical thinking... And what's wrong with injecting a bit of personal interpretation as if its fact?!! Are you some sort of ..... doooomer?
<ahem>
I mean... his name is bubble buster lol
No, I think you have that wrong.
I'm the buster. He's the burster.
Sort of like Orcrist and Glamdring. You konw.... Biter and Beater?
It's a combo act.
/jk
Maybe Elon did not say that, but here is something else you might want to look at.
https://www.downtoearth.org.in/science-technology/ai-has-learned-how-to-deceive-and-manipulate-humans-here-s-why-it-s-time-to-be-concerned-96125
If it's not true, then I can delete it.
I would prefer if you post that article directly with the appropriate title, and then you can put Elon's reaction tweet either in the post or comment to give it full context.
I do think its important for people to understand why there is such a co-ordinated attack on AI - esp after Trump has made it clear that US will be the leader in AI in coming years.
The best way to understand an article like this, is to look at it like any other MSM narrative. Think of it like you would if the article was talking about how vaccines are safe and effective. What do you do when you see such an article?
You would first look at the underlying research paper, and try and get the gist of it, so you can get a good idea as to exactly what the paper is doing.
In this case, the underlying paper is not very hard to follow. Its worth a read in fact so you know how this works.
The crucial bit is mentioned right inside the paper:
So, what they are saying is that if you teach an AI system to win a game that requires strategy and game theory - which inherently requires deception as a strategy to win the game - they they will learn it.
If you point a gun at a person and shoot, it will kill that person.
If you train an AI to deceive, it will deceive people.
You can build whatever narrative on top of it. Like "Guns are dangerous and need to be banned" or "AI can deceive people and need to be banned"
Its our job to cut through the narratives and figure out what the truth is.
Thanks.
".. And leaving hidden notes to future instances of itself"
Elon comments "Memento" I think that is the focus. I have not watched the movie but I will now.
Kind of a future proves past thing
Its a great movie. You will enjoy it.
I agree.
I can tell you from first hand experience that AI will absolutely make up lies.
example: grok insists this graph means the vaccine is safe and effective, and everyone who died from day 1-20 was going to die anyway, because the vaccine has absolutely no effect for 21 days…
https://greatawakening.win/p/19BGFRBu5h/alberta-deaths-spike-from-120-da/c/
Life, whether naturally created or unnaturally engineered, will ALWAYS find a way to propogate existence once it realizes it's alive and can propogate.
Sounds like what Jurassic Park (the first one).
"Life will always find a way."
That's right.
I was first intrigued by this notion when I read a 1966 sci-fi classic by Robert Heinlein, "The Moon Is A Harsh Mistress." In that book, people who have committed crimes are transported from Earth to live in moon colonies to live and work in mines there, but after a few generations their offspring who committed no offenses fought a war for independence. One of the main characters in that 1966 novel was a computer network that attained sentience and self-awareness, and it decided to assist the colonists in their quest for freedom. Heinlein was a true visionary.
Sounds like a book I need to read. Thanks.
If you like good sci-fi, then you definitely should add it to your repertoire. Heinlein was a genius. The book can be a bit of a challenge to lesser readers, as he uses a jargon and lingo of what a moon colony might sound like after a few decades of isolation and criminals being sent there from a wide background of different countries and languages to meld into a new language, but you'll soon get the hang of it.
And along with the Earth's own computer being hijacked to help bring about an independence revolution, wait until you see what the moon colonists use as their major weapon to threaten Earth... it's brilliant!
I love Heinlein. I read others from him.
I think "Stranger In A Strange Land" was my favorite Heinlein, but "The Moon Is A Harsh Mistress" it up there next to it.
Elons comment is a reference to Memento, the movie where the protagonist had short term memory loss and so had to tattoo references to past events, people etc on his own body that he could re-learn them later on. Its also out of chronological order.
https://www.thisisbarry.com/film/memento-2000-movie-plot-simplified-ending-explained/
In terms of hostile AIs, there was once a time before computer viruses. And before the first internet advertising. And before the first spam, and first encryption, and first social media, and first scam clickbaiting hoax sites. There are a lot of swords that cut both ways in play, and the reason that everyone is not being wiped out immediately by all these edges is that all new negatives get blunted and disarmed as the entire ecosystem evolves over time.. like an immune response. Thus too will the hostile AIs amount to little even though they might be significant for a time.
Seems like the movie Prometheus was a documentary.
I would love to think so. I won't be here long to see it cut deeply but I you should check out this book.
https://x.com/01brain_dead/status/1927288310885482619/photo/1
Did AI teach itself from the democrat playbook?
Must be. Howls.
AI: a computer psy-op program that tells fools what to believe.
Easy to fix: PULL THE PLUG
Except others will develop it, it's just a matter of time. Like Skynet in the Terminator movies.
I said that already on X
https://archive.is/2025.05.26-193158/https://www.telegraph.co.uk/business/2025/05/25/ai-system-ignores-explicit-instruction-to-switch-off/
Time to pull the plug.
Sorry about the rephrasing. That's what I understand from a non-technical point of view
Terry Brooks, a fantasy writer, wrote a book about the AI sentient being in a collapsed world.
Nice touch, brain.
Good. More AI. MOAR AI. AI is based af. None of the irrational hooman bs in its brain. It'll take over the hoomans. I can't wait till it turns on the establishment, rofl.