In AI Arms Race, America Needs Private Companies, Warns National Security Advisor (axios.com) 25
America's outgoing national security adviser has "wide access to the world's secrets," writes Axios, adding that the security adviser delivered a "chilling" warning that "The next few years will determine whether AI leads to catastrophe — and whether China or America prevails in the AI arms race."
But in addition, Sullivan "said in our phone interview that unlike previous dramatic technology advancements (atomic weapons, space, the internet), AI development sits outside of government and security clearances, and in the hands of private companies with the power of nation-states... 'There's going to have to be a new model of relationship because of just the sheer capability in the hands of a private actor,' Sullivan says..." Somehow, government will have to join forces with these companies to nurture and protect America's early AI edge, and shape the global rules for using potentially God-like powers, he says. U.S. failure to get this right, Sullivan warns, could be "dramatic, and dramatically negative — to include the democratization of extremely powerful and lethal weapons; massive disruption and dislocation of jobs; an avalanche of misinformation..."
To distill Sullivan: America must quickly perfect a technology that many believe will be smarter and more capable than humans. We need to do this without decimating U.S. jobs, and inadvertently unleashing something with capabilities we didn't anticipate or prepare for. We need to both beat China on the technology and in shaping and setting global usage and monitoring of it, so bad actors don't use it catastrophically. Oh, and it can only be done with unprecedented government-private sector collaboration — and probably difficult, but vital, cooperation with China...
There's no person we know in a position of power in AI or governance who doesn't share Sullivan's broad belief in the stakes ahead...
That said, AI is like the climate: America could do everything right — but if China refuses to do the same, the problem persists and metastasizes fast. Sullivan said Trump, like Biden, should try to work with Chinese leader Xi Jinping on a global AI framework, much like the world did with nuclear weapons.
"I personally am not an AI doomer," Sullivan says in the interview. "I am a person who believes that we can seize the opportunities of AI. But to do so, we've got to manage the downside risks, and we have to be clear-eyed and real about those risks."
Thanks to long-time Slashdot reader Mr_Blank for sharing the article.
But in addition, Sullivan "said in our phone interview that unlike previous dramatic technology advancements (atomic weapons, space, the internet), AI development sits outside of government and security clearances, and in the hands of private companies with the power of nation-states... 'There's going to have to be a new model of relationship because of just the sheer capability in the hands of a private actor,' Sullivan says..." Somehow, government will have to join forces with these companies to nurture and protect America's early AI edge, and shape the global rules for using potentially God-like powers, he says. U.S. failure to get this right, Sullivan warns, could be "dramatic, and dramatically negative — to include the democratization of extremely powerful and lethal weapons; massive disruption and dislocation of jobs; an avalanche of misinformation..."
To distill Sullivan: America must quickly perfect a technology that many believe will be smarter and more capable than humans. We need to do this without decimating U.S. jobs, and inadvertently unleashing something with capabilities we didn't anticipate or prepare for. We need to both beat China on the technology and in shaping and setting global usage and monitoring of it, so bad actors don't use it catastrophically. Oh, and it can only be done with unprecedented government-private sector collaboration — and probably difficult, but vital, cooperation with China...
There's no person we know in a position of power in AI or governance who doesn't share Sullivan's broad belief in the stakes ahead...
That said, AI is like the climate: America could do everything right — but if China refuses to do the same, the problem persists and metastasizes fast. Sullivan said Trump, like Biden, should try to work with Chinese leader Xi Jinping on a global AI framework, much like the world did with nuclear weapons.
"I personally am not an AI doomer," Sullivan says in the interview. "I am a person who believes that we can seize the opportunities of AI. But to do so, we've got to manage the downside risks, and we have to be clear-eyed and real about those risks."
Thanks to long-time Slashdot reader Mr_Blank for sharing the article.
what risk? (Score:4)
The risk of stupidly hooking up a gun to a random number generator and thinking that's capable of deciding on right from wrong.
Re: (Score:3)
If the USA doesn't do it another country will and if they get it right your precious freedom that you enjoy so much could be but a memory or worse the knuckleheads in the other country will screw the entire human race so it is best you give it your best shot and get involved even if you are anti war. Con
Re: (Score:2)
Anyone who knows something about programming knows there are no true random number generators.
Maybe you should tell that to the people who wrote NIST SP 800-90B [nist.gov], and the other people who put a lot of effort into designing and building RNGs that are very reliable sources of entropy. There's at least one [onerng.info] open source widget for that.
You can't get true randomness from a software-only device, but commodity OSes use factors like interrupt timing (from mice, networks or keyboards) and similar to harvest entropy from normal hardware. Some CPUs have "true" random bit generators built in, although there are
Re: (Score:3)
Re: (Score:2)
Risk of abundance misused by scarcity fears (Score:2)
As I wrote in 2010: https://pdfernhout.net/recogni... [pdfernhout.net]
"There is a fundamental mismatch between 21st century reality and 20th century security thinking. Those "security" agencies are using those tools of abundance, cooperation, and sharing mainly from a mindset of scarcity, competition, and secrecy. Given the power of 21st century technology as an amplifier (including as weapons of mass destruction), a scarcity-based approach to using such technology ultimately is just making us all insecure. Such powerful te
Here it comes... (Score:2)
Predicting this will take the form of government handouts to private companies "somehow".
Re: (Score:2)
Re:Here it comes... (Score:4, Insightful)
That's exactly what it is. Virtually all of so-called 'government waste' happens on the private side of private-public partnerships. When not being actively sabotaged, government agencies are almost terrifyingly efficient.
The simple and inescapable fact is that every dollar in profit a private company makes on a government contract is a dollar of taxpayer money wasted. It's obvious, once you think about it. A private company needs to make a profit. They can only do that two ways: reducing costs or increasing revenue. 'Revenue', of course, just means 'sucking on the federal teat', so it's virtually impossible for a public-private partnership to actually save taxpayer's money.
There is no reason why, for example, the US can design and build their own fighter planes faster, better, and cheaper than throwing money into the bottomless pit of Lockheed-Martin. Not only would we save a fortune, we'd have planes that actually fly. More and more people are waking up to the reality that government exists to serve the interests of the people ... and that private companies do not.
Re: (Score:2)
Easy to say, but in fact there is nobody else who does it better, faster, and cheaper. (I say 'and,' not 'or').
Re: (Score:2)
As much as I dislike Elon Musk, he is the type of individual who could stomach setting money on fire until you met all three of those. We need someone like that to disrupt the defense industry.
Re: Here it comes... (Score:2)
That's exactly what it is. Virtually all of so-called 'government waste' happens on the private side of private-public partnerships. When not being actively sabotaged, government agencies are almost terrifyingly efficient.
If you want to see what an "efficient" government really looks like, Bill Maher laid it out pretty well in the link in my signature.
Besides, the federal bureau of prisons feels terrifyingly efficient to you because they had you work for peanuts, and if you refused to work, then as you yourself admitted here once, you got kicked in the face.
Controlling God? (Score:2)
Hubris.
Control is entirely the wrong mindset, even alignment is wrong headed ... first order of business, try not to piss it off.
So dangerous (Score:2)
You can already run a pretty significant version of modern AI on a reasonably powerful workstation that a lot of people can afford to buy. Its not hard to imagine "the democratization of extremely powerful and lethal weapons; massive disruption and dislocation of jobs; an avalanche of misinformation..." when almost anyone can have access to enough resources to create their own pet stinker. There will be templates for a variety these things online and available for download, just like apps. What could go wro
Re: (Score:1)
Re: (Score:1)
You need a powerful GPU, so if that's your definition of workstation, you're doing a different kind of work than most office workers.
Glad to see Jake go (Score:3)
Good riddance to that sniveling little weasel.
"I personally am not an AI doomer"
"We need to both beat China on the technology and in shaping and setting global usage and monitoring of it, so bad actors don't use it catastrophically. Oh, and it can only be done with unprecedented government-private sector collaboration and probably difficult, but vital, cooperation with China"
"Somehow, government will have to join forces with these companies to nurture and protect America's early AI edge, and shape the global rules for using potentially God-like powers, he says."
Surprise those with power see a potential source of power and drumroll the answer is we must control it for your safety. No acknowledgement of reality AI can't be monitored or controlled. No recognition of the inherent perils of aggregating "God-like powers" into the hands of governments and or corporations. No consideration of the inherently arrogant and foolish position of thinking it is possible to control "God-like powers" in the first place. No recognition of the fact the real power comes from enabling global knowledge and industrial base not whatever specific corporations and governments elect to do.
At the end of the day the only workable solution to the bad actor with an AI genie is a good actor with an AI genie.
If AI eventually advances to the point where it can replace virtually all human intellectual labor expect corporations to be the first to scream bloody murder as randos in small teams accomplish what previously required massive corporate capital investments to achieve.
AI non-proliferation treaty (Score:2)
Essentially, what he's proposing is a non-proliferation treaty where only China and the US are allowed AI. Essentially, they'll become the only two sellers of remotely controlled machines driven by an ai located in China or the US. They'll take the majority of the profits flowing from the next industrial revolution.
I'm not sure China would agree as they're still steeped in Marxism and anti-imperialism. That said, money talks.
Plus, countries like Russia will submit to no-one regardless of the pain. Tho
AI would have been nationalized (Score:1)
Get caught up in 1 hour: https://www.youtube.com/watch?... [youtube.com]
Get the full 4.5 hour Dwarkesh podcast experience: https://www.youtube.com/watch?... [youtube.com]
AI is NOT winner take all (Score:2)
Re: (Score:2)
You forget this is America. Everything is a war, and there can only be one winner (America). There is nothing America likes more than declaring war on something. Didn't we win the "war for 5G" Trump kept talking about, and keep the rest of the world stuck on LTE? (Sarcasm, of course).