So where's the guardrails?

Status
Not open for further replies.
Thank God we have members of PNWFF willing to parrot doom scenarios that we have already all read about and watched content from common news resources outside of this fly fishing website. Truly heroic work!
thnx, can tell from your snipe barrage what a fan you are
Feel free to look at the recent history or even the present of communist China and tell me they are on par. Slavery. Harsh treatment of native Americans and the treatment of returning vets from Vietnam, the three great American atrocities don't even come close to comparing to China . Just under Mao a death toll of 55 million. The great leap forward they called it.
Nothing America has done even comes close.
You want American Ai not Chinese even if you won't acknowledge it. It really is a one or the other choice. Just like the cold war, we have to win this.
Agree with you, Rob. China is determined to displace the US as the worlds Superpower and it will use AI at every opportunity to pull every lever to do so. Regarding their efforts with AI, without the access to Nvida chips and our brains, they have developed AI models they are selling that have less horsepower for one tenth the cost. These AI agents are now being sold around the world in many 'third world' countries, and who knows what backdoors for Chinese control come with them.
One of the accepted consequences of powerful AI's is the 'ours against yours' model, in which one countries AI's are constantly probing other countries infrastructure for infiltration weaknesses...water systems, power grids, banking, defense, etc.
This has already happened with cyber attacks on water systems in Minnesota, Georgia, New Jersey, Arkansas, and Michigan. Add Russia, North Korea and Iran to the list of known villains in this.
The real fear, however, is these AI agents become so powerful that they act independent on their own behalf.
One of the core challenges is where one would think a planted charge triggered at a data center complex would end an out of control AI, it's been established that high end AI agents may have the ability to transfer nodes of itself through the internet to computer complexes elsewhere, creating a distributed network it can operate from, even if it's original 'mothership' is destroyed.
Interesting times we live in.
 
You really don't understand what AI is do you ?

There's already models from dozens of countries, and it's not like the world is going to adopt one standard AI model, and then dispose of all the rest.
It's like saying 'you want American nuclear bombs, not Russian nuclear bombs', in the 50's, and then spend the next 30 years claiming your bombs are better.. when everybody knows both sides have them and they're destructive if used.

Every country that is developing AI will have a version of AI, and those that are not developing AI currently will just buy, steal or partner with a country that has it.
I read somewhere that once one AI model is advanced enough to improve it's own code, no other model will be able to catch up. Winner take all. I'm not sure if that's true, but it is maybe a difference with bombs and other technology.
 
I read somewhere that once one AI model is advanced enough to improve it's own code, no other model will be able to catch up. Winner take all. I'm not sure if that's true, but it is maybe a difference with bombs and other technology.
Not sure on that, as hardware is the bottleneck for AI improvement right now I read somewhere.
Like all exploration, nobody knows how things will end up, or what will be found.
 
As I was reading in an article I recently wrote, it will soon be impossible to protect our assets from theft by AI. This is why I'm converting all my credit union savings, retirement saving, and mutual funds to quality fly fishing equipment. When all my assets are tangible material goods of high value and entirely unplugged from the internet, I expect to be safe from any evil intent of AI.
 
As I was reading in an article I recently wrote, it will soon be impossible to protect our assets from theft by AI. This is why I'm converting all my credit union savings, retirement saving, and mutual funds to quality fly fishing equipment. When all my assets are tangible material goods of high value and entirely unplugged from the internet, I expect to be safe from any evil intent of AI.
Bullets, whiskey, & fly rods- Salmo G
 
AI hookers and blow...
That's when the technology will peak...

The actual issue with AI from where I see it, is AI does not know right from wrong. It follows data rules rather than an ethical compass. Theres no intrinsic values system as humans know a values system. The idea that one country's values would somehow be integrated, or a particular religion be the AI's lodestone..Will there be Buddhist AI, a Sikh AI, a Scientology AI ?

But hey... look at all the money there is to be made...so... never mind.
 
AI hookers and blow...
That's when the technology will peak...

The actual issue with AI from where I see it, is AI does not know right from wrong. It follows data rules rather than an ethical compass. Theres no intrinsic values system as humans know a values system. The idea that one country's values would somehow be integrated, or a particular religion be the AI's lodestone..Will there be Buddhist AI, a Sikh AI, a Scientology AI ?

But hey... look at all the money there is to be made...so... never mind.
They call this the alignment problem.

Makes it sound less bad, I s'pose.
 
Yup...

A virtual soul isn't actually a soul. Just like virtual hookers and blow isn't actually hookers and blow.

😁
 
AI hookers and blow...
That's when the technology will peak...

The actual issue with AI from where I see it, is AI does not know right from wrong. It follows data rules rather than an ethical compass. Theres no intrinsic values system as humans know a values system. The idea that one country's values would somehow be integrated, or a particular religion be the AI's lodestone..Will there be Buddhist AI, a Sikh AI, a Scientology AI ?

But hey... look at all the money there is to be made...so... never mind.
I'm not quite so ready to dismiss the source of implicit values. It's a word predictor, and the training data will have biases based on the culture and language used.

In other news, I got a flyer in my email about an upcoming TU meeting where data centers and water quality is one of the agenda items. The flyer was clearly produced using AI. :confused:
 
An interesting discussion

 

Did everyone (Rob) realize that US was way out in front on this? I find it interesting that China, a country where there is little protesting allowed, is so far behind the US where there has been a ton of protesting???? Strictly anecdotal on my part but local folks here in the NCW are pissed off (both right and left) about data centers but all cry a common theme that commissioners are getting $$$ petted in advance. Thus by the time people catch on, it's too late!
 
On advice of my son we redistributed our retirement funds to a half dozen accounts as a safeguard, use .
VPN's and password generator Lastpass, and delete internet memory and scrub cookies daily.
Son was returning home from a London business trip earlier in the year when his phone, which has heavy security software installed, alerted him it was being hacked so out came the battery and SIM card.
Meanwhile, every day there is more...

"AI threats that sound like science fiction are much closer to today’s reality, Anthropic warned today — and some may already be here, POLITICO’s John Sakellariadis reports. “In a lengthy new report out Thursday, Anthropic said that criminal hacking gangs, Chinese security bureaus, Russia-linked spies and Yemen-based arms manufacturers, among others, have exploited a web of fraudulent accounts and online services to access some of the company’s advanced AI assistants, known as Claude. And they’re using those tools to attempt attacks or design weapons that previously lay beyond their capabilities.”
The company also said it “disrupted several potential plots this year by scientists who used its leading artificial intelligence models to conduct research that could have helped develop biological weapons,” per NYT’s Dustin Volz.
Meanwhile: OpenAI held meetings with top power companies to talk about securing the electrical grid, as the company’s products become embroiled in cyberattacks. CEO Sam Altman met with energy execs at Edison Electric Institute’s meeting in Colorado yesterday to offer OpenAI’s cyber services for protections, but it comes following revelations that AI platform Hugging Face was hacked by hundreds of AI agents, which has raised alarms about possible attacks on critical infrastructure.
On the Hill: Sen. Josh Hawley (R-Mo.), who chairs Senate Homeland Security’s Disaster Management subcommittee, is investigating OpenAI’s handling of the Hugging Face breach and demanding a response from Altman by Oct. 1, Axios’ Andrew Solender and Maria Curi report. And the Senate Commerce Committee could take up AI regulation when the chamber comes back next week, Sen. Ted Cruz (R-Texas) told Playbook’s Dasha Burns in the latest episode of “The Conversation.”
 
thnx, can tell from your snipe barrage what a fan you are

Agree with you, Rob. China is determined to displace the US as the worlds Superpower and it will use AI at every opportunity to pull every lever to do so. Regarding their efforts with AI, without the access to Nvida chips and our brains, they have developed AI models they are selling that have less horsepower for one tenth the cost. These AI agents are now being sold around the world in many 'third world' countries, and who knows what backdoors for Chinese control come with them.
One of the accepted consequences of powerful AI's is the 'ours against yours' model, in which one countries AI's are constantly probing other countries infrastructure for infiltration weaknesses...water systems, power grids, banking, defense, etc.
This has already happened with cyber attacks on water systems in Minnesota, Georgia, New Jersey, Arkansas, and Michigan. Add Russia, North Korea and Iran to the list of known villains in this.
The real fear, however, is these AI agents become so powerful that they act independent on their own behalf.
One of the core challenges is where one would think a planted charge triggered at a data center complex would end an out of control AI, it's been established that high end AI agents may have the ability to transfer nodes of itself through the internet to computer complexes elsewhere, creating a distributed network it can operate from, even if it's original 'mothership' is destroyed.
Interesting times we live in.
So you use the principles Schwartzkopf and Powell used in Desert Storm. ("1st We're going to cut it off. Then we're going to kill it.")
Details up to you.
 
Last edited:
Status
Not open for further replies.
Back
Top