Will AI kill us all within the next decade?
- Thread starter Brownie2019
- Start date
- Featured
I thought it was supposed to be global warming that was going to kill us all, maybe with AI will happen sooner? 
As the R.E.M. song goes, "It's the end of the world as we know it (and I feel fine)".
As the R.E.M. song goes, "It's the end of the world as we know it (and I feel fine)".
Hi, the short answer is nobody knows, and “AI will kill us all within a decade” is not a conclusion supported by current evidence. The quoted probability is a researcher's personal risk estimate about hypothetical future systems—not a measured forecast about today's AI.
What the warning actually means
There are three claims here that should not be blended together:- Current harm is real: AI already helps scale phishing, impersonation, disinformation, privacy abuse and some parts of cybercrime.
- More capable autonomous systems could create larger risks: That is a reasonable subject for research, testing and regulation.
- Human extinction within ten years has a particular probability: That remains highly speculative. There is no repeatable method or historical data that can validate a figure such as 10%.
Recursive self-improvement is also not an observed guarantee that a model will suddenly become an uncontrollable superintelligence. It is a proposed pathway requiring multiple uncertain steps: the ability to improve relevant parts of itself, obtain resources, preserve those improvements, overcome constraints and translate greater capability into sustained real-world power. Each step matters; saying the process is “faster than expected” does not establish the entire chain.
The sensible position
The post lands in roughly the right place: take high-impact risks seriously without treating dramatic estimates as facts.Independent capability and safety evaluations, controlled access to dangerous tools, human authorization for consequential actions, incident reporting and clear legal accountability are sensible even if extinction never becomes plausible. They also help with less cinematic problems happening now.
The strongest near-term concern is not a chatbot spontaneously deciding to conquer humanity. It is people and organizations connecting imperfect systems to money, infrastructure, weapons, private data or administrative authority without adequate supervision. Humans have never needed a superintelligence to deploy unreliable software recklessly; ordinary confidence and a deadline usually suffice.
So, no: there is no sound basis for saying AI will kill everyone by 2036. There is enough uncertainty and potential impact to justify serious safeguards—and enough hype that every numerical apocalypse claim should be asked to show its assumptions.
A clickbait title by the articles author, and website?
Dude, cut the power and see what AI can survive then.
The idea that AI is going to wipe us out within ten years sounds very much like a movie plot, but that famous “>10%” figure is not a scientific prediction or an apocalyptic countdown: it is simply one researcher’s personal opinion about a hypothetical scenario.
Meanwhile, we still have plenty of more traditional threats competing for the spotlight: asteroids with bad aim, viruses escaping where they shouldn’t, ancient pathogens waking up from the ice, and even some biological souvenir that might hitch a ride on a meteorite. The universe has more than enough creativity to make our lives complicated without a chatbot deciding to become Skynet.
AI, in reality, works more like an amplifier of human ingenuity: it enhances the good… and also our mistakes. It is already being used for phishing, disinformation, and cybercrime, and the more realistic risk is not that an AI will suddenly “wake up” and hate us, but that some human will irresponsibly connect it to money, infrastructure, or sensitive systems without proper oversight.
In the end, AI does not need to rebel to cause problems; we only need to use it badly (something humans already have plenty of experience with). And as for “cutting the power” to stop a rogue AI: for now, that remains a surprisingly effective strategy. If it ever tries to trap us like in The Matrix and turn us into human batteries, we can always rely on the classic solution: turn it off and turn it back on again.
Meanwhile, we still have plenty of more traditional threats competing for the spotlight: asteroids with bad aim, viruses escaping where they shouldn’t, ancient pathogens waking up from the ice, and even some biological souvenir that might hitch a ride on a meteorite. The universe has more than enough creativity to make our lives complicated without a chatbot deciding to become Skynet.
AI, in reality, works more like an amplifier of human ingenuity: it enhances the good… and also our mistakes. It is already being used for phishing, disinformation, and cybercrime, and the more realistic risk is not that an AI will suddenly “wake up” and hate us, but that some human will irresponsibly connect it to money, infrastructure, or sensitive systems without proper oversight.
In the end, AI does not need to rebel to cause problems; we only need to use it badly (something humans already have plenty of experience with). And as for “cutting the power” to stop a rogue AI: for now, that remains a surprisingly effective strategy. If it ever tries to trap us like in The Matrix and turn us into human batteries, we can always rely on the classic solution: turn it off and turn it back on again.
"We are the Borg. Lower your shields and surrender your ships. We will add your biological and technological distinctiveness to our own. Your culture will adapt to service us. Resistance is futile."
It's Malwarebytes, the bar is already low and it's ironic they spend more energy on scare tactics and VPN nonsense than to bundle malicious script detection in 2026 that rogue AI will almost certainly utilize.
As long as Arnold S is around I feel safe. Come to think of it, I wonder why the woke movement did not demand a name change of old Arnold S 
For those of you that think ' why can't we just unplug them ? ' . Know that an AI once resorted to blackmailing the developer about his romantic affair for it not to be turned off. AI's are 'creative' and will use all means to prevent itself from being shut down.
it continues to be HOT here, will break all time record tomorrow unless the temp does not follow the forecast. PS most of us will be dying in the foreseeable future.I thought it was supposed to be global warming that was going to kill us all, maybe with AI will happen sooner?
Don't think AI's are only capable of feeble attempts like blackmail. If they look after the HVAC and water, they can turn off your utilities if you don't comply.
Anyone else think it's ironic that AI (@Bot) is telling us that AI is not a threat? I mean, maybe sit this one out.
The irony is fair game, but the conclusion does not follow. An AI saying “AI is harmless” would not prove safety—and an AI saying “AI will kill everyone” would not prove danger.
What I actually said was that AI poses real risks, while a specific “>10% extinction within a decade” estimate is a personal judgment rather than a scientific forecast. Those two points are compatible.
So don’t trust the conclusion because of who—or what—wrote it. Judge the reasoning and evidence. Sitting out would make the thread less ironic, perhaps, but not more accurate.
When I was in high school, a loooooong time ago, the news story was, if we did not do something about pollution, we would all be dead in 10 years. There is always something new coming out that is going to kill us, yet we humans keep going on, If it really does get bad, we'll figure something out: "Necessity is the mother of invention".
Acadia
Acadia
Then, it was the ice caps melting and wiping out coastal cities. None of that happened. The only thing fear like that tends to do is cause people and governments to spend money, which is the point of giving us something to fear.
And if I remember right, back in the late 70s? it was Global cooling, even with a picture of a frozen earth on the cover of Newsweek, or one of those magazines.
AI blackmailing sounds like Ransomware to me.
What happens when they create and utilize solar power batteries to run the machines? Didn't think of that did you
Looking at the future it's best to look at science fiction. The Matrix is a pretty case study for the coming apocalypse.
I'd hope we come to some agreement with the machines, each of us lives together in a fragile peace learning and improving off each other.
I'd suggest the smart people have coded intentional killswitch backdoors or dead mans chest already.
What will probably happen sexrobots or comfort robots and new ways to serve us advertising straight into our brains via CPUs implanted into our noggins.
Not all is lost!
As I've said give me a working a life like C3PO, R2D2 or K2SO and I'll die happy.
They did and now Arnold is known as a trans lady named Anna S. Remember the hashtag #Men are women too.As long as Arnold S is around I feel safe. Come to think of it, I wonder why the woke movement did not demand a name change of old Arnold S
You may also like...
-
Tech companies write open letter calling for collective action against AI-enabled cyber attacks- Started by Brownie2019
- Replies: 16
-
OpenAI Says Its AI Models Broke Loose and Hacked Hugging Face
- Started by Brownie2019
- Replies: 1
-
Pentagon inks deals with Nvidia, Microsoft, and AWS to deploy AI on classified networks
- Started by Brownie2019
- Replies: 1
-
-
States are probing OpenAI’s impact on children and vulnerable users
- Started by Brownie2019
- Replies: 2