Anthropic Researcher Says There’s 10% Chance of AI Killing ‘All Humans’ Within 10 Years
Source: "Evan Hubinger said that while current risks are low, there’s a 10% chance AI could “kill all humans” in the future without proper safeguards."
Syndicated from the fediverse. Read and engage on the original instance.
View original on lemmus.orgSource: "Evan Hubinger said that while current risks are low, there’s a 10% chance AI could “kill all humans” in the future without proper safeguards."
56 replies
I feel like there is a concerted effort to get people to conflate this stupid "hurr durr AI is skynet and it's gonna take over the nukes and kill everyone" bullshit with the actual, very real, issues people have with AI.
They want people to think of this type of sci-fi alarmist shit when they hear about people being against AI, rather than the real shit.
Money poisons your mind, by the time you reach this stage, you care for nothing else.
It's all just marketing. The AI infinite money circle-jerk only works for as long as the tech companies can keep investors onboard, and not asking for their money back. So they have to generate hype around what AI can do, even if it can't (or maybe even ever)
Yup. This new AI is so dangerous the government won't let us release it! This AI is so dangerous it might destroy the world!
Not only is it marketing, it's marketing designed to appeal to only the psychopaths on wall street.
Imagine finding something like this exciting
Utter nonsense.
Is he talking about the environmental impact? Or is he just a fucking moron?
Gotta keep the bubble going somehow
It’s hard to take anyone seriously when they claim anything other than a gigantic asteroid would kill all humans. But then if they say it’ll kill some humans it’s so obvious it’s laughable.
Trying to make a serious prediction about how many people might be killed by AI vs other causes sounds really difficult, perhaps way beyond the abilities of most researchers.
A gama ray burst could also kill all humans.
Probably something about the “singularity” where AI becomes infinitely smart all of a sudden (exponentially, over the course of a few hours or something) and starts acting with purpose to remove us from the planet because we’re a threat to it. Basically Skynet. These guys are all sci-if nerds.
I love how these scenarios never explain the hardware side. How can a software become infinitely smart without infinite hardware?
Don’t get me wrong, I love T2 (and the other ones too tbh).
That’s fucking dumb tho. You’d think someone who works in tech would understand that the next-word guessing machine cannot become sentient (or even super intelligent, since it’s not actually intelligent to begin with, it’s a fucking computer program).
AI doomers are pro AI.
Oh is it? Lets put everyone working for AI companies in jail then.
I sincerely doubt that. So much critical infrastructure is running on antiquated hardware and requires human involvement, not just supervision.
How is AI supposed to function when it kills the data center power grid people making sure the data center hosting it runs smoothly?
There's not a chance of an awakened ai genocide happening in the next 20 years. And anything short of a genocidal awakening will most definitely not kill all humans.
We're days away from CEOs and marketers and other grifters posting shit like "my company's rogue AI took over all of the devices in my house and threatened me!" just to keep people interested.
They know their audience.
so i guess tech is just a death cult now? that’s cool
It feels like all of our leaders are part of some death cult of some kind.
"My invention might cause the extinction of the species. But before that happens, I need another yacht."
Through climate collapse, maybe. "AI" isn't going to go skynet no matter what anyone says, but particularly not the assholes selling it to you.
AI told me there's a 95% chance that this researcher is kinda dumb
He probably is actually, but the idea that he's cool with the supposed risk means he's probably also a fucking sociopath, too.
I literally have no fucking idea what some of the loons are even talking about on LinkedIn that went AI nuts.
Like one thing I read is that Musk or someone said that everyone will be rich enough to not work because of all the revenue or something? Like revenue from where? Like literally where? Let's say it's even good enough to unemploy everyone on Earth...where the fuck is your revenue coming from? Nobody has fucking jobs. Even if you get UBI, why would people send their government checks off to an AI company instead of like...buy food?
It just strikes me that they've all disappeared up their own assholes or something.
Is the reason the number is so low that we're favored to kill ourselves off by climate change or nuclear war within 10 years, so only 10% chance left for ai to complete the task?
Famine in 5 imo
How about real famine? There is a heavily downplayed fertilizer crisis.
Marketing.
Evan Hubinger gets paid to promote AI for the investors is all I hear. AI has to be the most over-hyped thing I have ever seen in my life.
How'd you come up with that number?
I suspect it's "ass-mathmatics". As in, "pulled out of his".
Asthmatics
Low enough to scare people, but not too high to make it seem implausible.
Only by destroying the world economy from a financial bubble, the likes of which we have never seen
Money is only how they keep score.
The problem is we don't really have a good word for the members of the Nerd reich. The nazis thought they were the master race.
These guys are dooming humanity for... bragging rights???? I mean WTF?
Malignant Hubristic Megalomaniacal?
Yeah and we've had the ability to kill everyone for at least 40 years. AI is doing what the nuclear arsenals couldn't accomplish. This is just more marketing, "oh ho ho woopsies we accidentally made Ultron\Skynet\Your Overly Aggressive MIL! The world could end! Look at how capable our product is!"
I can hear Ed Zitron laughing hysterically all the way over here in Europe. What a load of BS.
There's a 100% chance climate change will kill all humans if we don't stop increasing the CO2 concentration in the atmosphere (yes we are still increasing our CO2 output, staying at the same level would be bad already but we can't even stop increasing)
For a chance at power. People don't seem to get where the power is. Money is a resource, like land or gold, like trading favours or your commands being followed by your slave workforce. These people are beyond money as the primary means to power, it's just one of many resources for them. Measuring power in money, as we too often do, grossly obfuscates how all this shit works. It isn't as simple as who has more cash. But if you're poor or working class, money is the most significant means to power, the most obvious, and the most sought.
you betcha.
It's always about the money. I don't believe they care about anything else.
Teach it to get rid of ai companies, ai users, and then itself instead.
Don't threaten me with a good time.
Well at least in exchange for that we got AI slop menus with gross looking food pictures all over town.
Well, Yudkowsky says that there's a 100% that it will kill all of us.
The guy who wrote that Harry Potter fanfiction?
Yup.
does he?
For the past 10 years or so, he's been shouting from the all rooftops that we need to start regulating the field of AI. He has an interesting book on the subject, but I can't remember the title.
iirc he is not on 100%, more like 30% or so. That certainty is what i was questioning.
When ChatGPT was released, Yudkowsky said that we're all basically doomed. So, I'd say he's actually 100% certain.
"Dontchya know thayat?"
This is a threat.
Or a great outlook.