In einem von der Regierung in Auftrag gegebenen Bericht heißt es, dass die USA „entschlossen“ vorgehen müssen, um eine vom Aussterben bedrohte Bedrohung durch KI abzuwenden
https://time.com/6898967/ai-extinction-national-security-risks-report/
In einem von der Regierung in Auftrag gegebenen Bericht heißt es, dass die USA „entschlossen“ vorgehen müssen, um eine vom Aussterben bedrohte Bedrohung durch KI abzuwenden
https://time.com/6898967/ai-extinction-national-security-risks-report/
35 Kommentare
„The U.S. government must move “quickly and decisively” to avert substantial national security risks stemming from artificial intelligence (AI) which could, in the worst case, cause an “extinction-level threat to the human species,” says a report commissioned by the U.S. government published on Monday.
“Current frontier AI development poses urgent and growing risks to national security,” the report, which TIME obtained ahead of its publication, says. “The rise of advanced AI and AGI [artificial general intelligence] has the potential to destabilize global security in ways reminiscent of the introduction of nuclear weapons.” AGI is a hypothetical technology that could perform most tasks at or above the level of a human. Such systems do not currently exist, but the leading AI labs are working toward them and many expect AGI to arrive within the next five years or less.
The three authors of the report worked on it for more than a year, speaking with more than 200 government employees, experts, and workers at frontier AI companies—like OpenAI, Google DeepMind, Anthropic and Meta— as part of their research. Accounts from some of those conversations paint a disturbing picture, suggesting that many AI safety workers inside cutting-edge labs are concerned about perverse incentives driving decisionmaking by the executives who control their companies.“
>U.S. Must Move ‘Decisively’ to Avert ‘Extinction-Level’ Threat From AI, Government-Commissioned Report Says
True.
Though, it will take a global ban. Hard to unilaterally withdraw when state- and non-state actors might press ahead.
This report is reportedly made by experts yet it conveys a misunderstanding about AI in general.
(edit: I made a mistake here. Happens lol. )
edit[ They do address this point, but it does undermine large portions of the report. Here’s an article demonstrating Sam’s opinion on scale [https://the-decoder.com/sam-altman-on-agi-scaling-large-language-models-is-not-enough/](https://the-decoder.com/sam-altman-on-agi-scaling-large-language-models-is-not-enough/) ]
Limiting the computing power to just above current models will do nothing to stop more powerful models from being created. As progress is made, less computational power will be needed to train these models.
Maybe making it so that you need a license to train AI technologies, punishable by a felony?
How exactly is AI going to make us go extinct? Like sure if Skynet becomes real but we’re so far from that it’s basically the equivalent of spears to nuclear weapons
Weird how these reports often boil down to “Give us funding or America is fucked!”
As someone who is in the AI-field, this is staight-up fearmongering at its finest.
Yes, AI is getting more powerful, but it’s nowhere near a threat to humans. LLM models lack critical thinking and creativity, and on top do hallucinate a lot. I can’t see them automating anything in the near future, not without rigorous supervision at least. Chat- or callbots sure, basic programming sure, stock photography sure. All of them don’t require any ceativity, at least in the way they’re used.
Even if these things are somehow magically solved, it still requires massive infra to handle huge AIs.
Also, they’re all GIGO until now – garbage in, garbage out. If you finetune them to be friendly, they will. Well, until someone jailbreaks them 😉
I think Humans are doing a pretty damn good job extincting much of what lives on this planet, and eventually ourselves. We don’t need the help of AI. Just look at what’s happening around Florida at the moment. Some severe mass die-offs are happening all around that state and scientists are horrified and scared of what they are seeing. Shit’s going to get real, real quick.
I’m sure this will be met with the same serious tone as reports about climate change.
I still dont understand what they think is going to happen. Terminator is a great movie but also far fetched. Cant imagine AI doing much else other than robbing people of various types of jobs. I also doubt we (or any other country) would just hand it the keys to nukes, cross our fingers, and go on vacation.
Well hopefully the A.I will be a less shit-tier civilization than we are I guess
Extinction for politicians and governments. That is all they’re concerned about.
Is the entire report just quoting Sam Altman’s fear mongering to try to get Congress to shut down his competitors again?
Fkn cops scared of the AGI supermind telling us no when we ask it if capitalism is good.
For the USA, decisive actions means actions that take years to reach instead of decades, a slowness resulting from the typical political in-fighting that goes on in a 2-party system.
As such, many big government measures in the USA are reactive instead of proactive, resulting in damage done instead of damage prevented.
Do we really have to use the term “extinction level threat” for everything? This is just fear mongering by people paid to write government reports. If they say, ai is no problem government won’t give them 1/4 million dollars to write the next report.
There should be legislation around AI, to protect people. But limiting computing power? What about China, or Russia? They will be where we are in no time. You can’t limit the raw power of AI, but you can agree that more characterization needs to be done with each generation of ai, so we can reap benefits and flag potential problems.
But we’re just going to keep twiddling our thumbs about the global warming extinction level threat? Ya know, the one that’s real, already happening and not vaguely theoretical. Bring on our robot overlords because they probably won’t be any worse than our human ones.
Here’s an idea: AI might eventually insert itself into the www and mess up all algorithms and search results, at the minimum. People worry that ‘xxxx’ country will control AI – no, once AI reaches that level no country will be in control.
Yeah this wont happen. You cant just stop this stuff in the US and think it will stop everywhere. Or that the world will somehow agree to do this. Just 100% unrealistic, and anyone suggesting this probably intends to find a loophole to it, or do it in another country.
I suspect this has something to do with diverting funding away from the two major active threats identified right now, because they’re „political“.
Problem is the cat is already out of the bag. It’s not like other state actors aren’t developing it for themselves. So sure the US can stop all development within it’s borders then have all of it’s systems pwnd by someone else’s super awesome AI and then succumb to autonomous machines in combat. Fighter jets, tanks, ships, drone swarms better and faster than any manned vehicle and with none of the human logistics like food, housing, and so on. At minimum the psyops cold war that’s been going on will be put into overdrive. A bomb never has to be dropped to destroy a country. So yeah I’m sure they will totally stop developing strong AI.
kind of confused, LITERALLY just read about the new semiautonomous defense systems for 2028 in the military with nonpilot aircraft.. sooo
There will be no action taken until people are revolting in the streets. Honestly, mass unemployment will hit eventually, and I just hope it’s bad and fast instead of a slow trickle by loss of jobs.
The sooner we get past this the better.
US Lawmakers are as suited to this task as rugby players are to international politics. This can only end badly.
pfffffttt, asking the US to do this?
I saw that we just passed the anniversary of eliminated leaded gasoline recently. If that was an issue today there is a basically 0% chance the US government would do anything about it. They barely blink an eye at lead water pipes still in use.
At this point we have so many “extinction level threats” that AI is just going to have to get in line and take a number.
Unlike concerns over climate change the probable time-line for this sort of thing is really rather rapid. People and governments can see it happening in front of them, bit by bit.
Fear mongering. They are worried about their monopoly on slave labor disguised as jobs and mortgages with never ending inflation.
We need to start protecting our libraries, as they will become the vital resource for our future.
Bullshit. They just want to monopolize control of AI.
If they were actually worried, they wouldn’t be experimenting with AI control of war zones, and mounting guns on robot dogs.
I’ll give you a real world example of how AI could actually be weaponized.
Let’s imagine a foreign government that designs AI to break into critical computer systems. The AI is programmed to detect the devices within your network and „brick“ all of them. Lets also imagine the AI does this with incredible speed to all our vulnerable computer systems nationwide. Everything goes offline at the same instant – power systems are offline because the AI tanked all the devices, internet offline, cell phones offline, satellites offline, all your files are erased – all backup files are erased.
How long would it take to get everything back online? Months?
Or, even more nefariously, the AI not only bricks your system but also sets things into motion to ruin them. Like setting a nuclear reactor to overload, then bricking all the computer systems – imagine if every nuclear plant in America was a Chernobyl like disaster. Setting electric power plants to ruin or explode. Giving wrong coordinates to planes, sending them crashing into the ground or each other. Just imagine the various ways a rival nation could weaponize AI – and you could imagine if AI got out of control and turned on every network (including those who tried to create it to attack their rivals, but instead started to attack their own systems).
We could be talking months, if not years of major disruption – including problems with food production and food distribution. You could have famine on your hands and riots breaking out worldwide.
This sentiment makes no sense to me especially after the comparison to nuclear weapons. How is it reasonable to expect the US to slow down development of AI if it’s powerful enough to destroy humanity? This would have to be worldwide agreement because if we don’t do it, someone else will.
It’s one of the many reasons the nuclear bomb was created.
A group of activist ideologues lobbied to receive $250k to write a “report” about their preexisting beliefs / concerns regarding AI safety. The reporting on this was irresponsible. Big Co is jockeying to erect castle walls around this new technology to make the barrier to entry harder for up-and-coming organizations. Classic pull up the ladder behind you.
here they go playing up yet another way to fear monger people.
I don’t know who I hate more, well….i know I hate the perpetrators who keep pushing this, but I find it hard not to hate the ignorant people who are falling for this, they will fight against people who know it’s bogus just so they can be a victim to a figment.
People watch a few science-fiction movies and think that AI is always going to be evil. The AI we have built has safeguards on top of safeguards to prevent it from doing what we don’t want it to do. Who did the government commission this report to?
This is ridiculous. If you knew what an actual AI looks like you would be dying seeing this as a headline. People don’t even know what they are talking about, quit Netflix already