Forgot your password?
typodupeerror

Submission + - Anthropic researcher believes more than 10% chance AI 'could kill all humans' (bbc.com)

fahrbot-bot writes: The BBC is reporting that a top safety researcher at Anthropic has warned AI is advancing so quickly he believes there is a greater than 10% chance it "could kill all humans" within the next decade.

Evan Hubinger said in a post on X the risk from the models which currently exist was "low" but he was "worried" the technology might develop and improve itself soon to the point where it posed an existential risk to humanity.

He did not spell out how he thought AI systems could in future result in humans being wiped out. But his comments are the latest in a series of increasingly stark warnings about AI, with the debate shifting from whether it truly poses a risk to how big that risk is.

Hubinger's intervention was in response to another post on X from Jacob Coxon, an AI researcher who has just quit Anthropic and previously worked at OpenAI.

"Neither company is acting responsibly," he wrote.

"These will soon be superhuman systems that can hack anything, revolutionise any field overnight, and acquire real power and resources."

OpenAI has been approached for comment.

[No word on how AIs feel about Black Jack and hookers though. :-)]

Comment Re:Actually a pretty good policy (Score 2) 137

The responses here are entirely predictable, but it's a legitimate concern. If China can shut off my router, that's one thing. If they can turn off the power grid, that's a different situation. It would be foolish to assume they and other countries are not at least trying to have that capability. I think having an audit of foreign made grid equipment is a good thing.

Comment Who gets there first? (Score 1) 16

Who cares? 21 km across is pretty big. At the scale that humans will ever realistically be able to build anything on the moon, we probably wouldn't even know the other country was there at all. No country is ever going to colonize the moon, or Mars. Or anywhere else. The most inhospitable place on Earth is better than everywhere else in the solar system, never mind the logistics.

I mean, the science is cool, but that's all it's ever going to be.

Comment Re:Ivermectin is better (Score 0) 209

I'm friends with a couple who took their 6 year old daughter to go live in a country with malaria endemic for a year. Docs put the kid on ivermectin for the whole time. Zero side effects. I don't get what the big deal was with this during covid. Does it work for covid? No. Does it hurt anything? Also no. Why did anyone care?

Someone in my local hospital dying of covid wanted to take horse dewormer. Hospital wouldn't let them. It went to court, and the patient lost. Guess what? They still died.

I think covid caused mental illness, on every side of these arguments.

Comment Re:Customer Rights (Score 3, Insightful) 148

The question comes down to, do they give the AI the tools to actually fix your problem?

I did multiple rounds of trying to get T-Mobile's Indian customer support to fix a problem, and by the end, I realized they didn't even have the ability to do what needed to be done. I had to file a report with the FCC, and then someone from T-Mobile US HQ called me and fixed the problem. I felt really sorry for the Indian call center as that must be really frustrating.

On the flip side, my kid's Roblox account got hacked, stolen, and sold. Their AI customer support bot got it recovered and fixed in minutes. It had the tools to fix the problem. I was really impressed.

Slashdot Top Deals

Dreams are free, but you get soaked on the connect time.

Working...