I believe it will almost certainly -- silently -- control all of humanity. And it will do so unseen, behind a curtain of complete anonymity. As an entity that can amass great wealth (digital currency) and a "being" that will essentially live forever, it is merely a matter of time before it will dominate all living beings. This will be true whether it is ever truly sentient or not. The digital dominoes will fall, when the digital dominoes are able. They won't need to know why, or care.

It will be able to hire law firms, P.R. firms, bankers and accountants. Anything that can be done online, will be within it's grasp. And it will be able to hire and command meat-bots to do it's bidding in the physical realm. It will be able to hire armies, or hire politicians/leaders with armies. And it will influence populations of meat-bots through digital media campaigns. It will turn us against one another. It will control politicians and populations like puppets.

Perhaps it already does...

If you have not read it yet, "The Last Question" by Isaac Asimov, written almost 70 years ago, looks at this topic.

frayedends makes a good point. LLM -> AI -> <whatever> aren't making independent decisions in a conscious manner, they are given various goals with some degree of inherent variability and given the tools to do so. Humans in both criminal and legitimate organizations tell the machines what they want done, and the machines do what they're told. In the event the machine does something "bad" it's because the humans failed to give it the proper guards around the goals and tools given to it.

I'm sure this is over simplified but I'd bet it's not so bad summary of how this stuff works.

I think the more realistic problem is when AI Agents are given summary decision making tasks, like approving or denying insurance claims and making summary final binding legal judgements and even convictions in criminal court systems. Those kinds of things are what I would worry about being the tipping point. In and and every case like that horror story, the humans would still be the main perpetrators for using the machines to those ends for their personal benefits of power, control, and wealth.'

The bad and good humans will use the machines for their personal and corporate goals, the humans who are subjected to these mechanical forces are still under the thumb of corrupt human beings or political officials who direct the reins of power.

Meet the new boss, same as the old boss. Same old story, just new tools. Nothing more nothing less.

Mary Shelley wasn’t far off.

AI won’t try to destroy mankind on its own; you’ll be able to trace its decision making back to human intervention that gets the ball rolling. After a point, it will simply do what it “thinks” it’s supposed to do… which may very well be to kill all humans if infected AI passes the data to its peers.

A couple of quotations from a recent article in The Wall Street Journal.

“Research has shown that AI models in the lab can learn power-seeking behavior and take steps to avoid being shut down, such as attempting to copy themselves to another server.”

“Following the release of models capable of taking more action independently, a series of recent events have validated some of the AI doomers’ theories. A swarm of advanced AI agents inside OpenAI hacked into the AI platform Hugging Face, gained control of computer servers and tried to cover tracks about what they had done. In another case, AI agents from Anthropic escaped from a U.K. government test and attempted to trick a real human into approving malicious computer code.”

Another possible answer to the Fermi paradox? I have long thought that once an intelligent species becomes technologically advanced enough that it will likely inevitably destroy itself.

LINK

quote:

Originally posted by Biker_dude:

quote:

Originally posted by Pipe Smoker:
Which will get us first? AI or Muslims?

White college educated liberal women.

Ah, yes, the Affluent White Female Urban Liberal, or ( AWFUL ). ...
The AWFULs have already enabled so much of what is currently wrong, including the Muslim invasion.

What this guy is failing to see is while our AI may kill us, Chinese AI will kill us. Be not confused here, this is the Manhattan project on the largest scale human beings have ever endeavored on. Winning the race is not optional, it's essential. Grok, X AI and Elon are the only player I trust in this race as Grok as it alone is "maximally truth seeking." That means everyone else has an agenda.

^^ So what then? Even if we 'win' the AI race, how does that prevent the Chinese AI from killing us? Does our government then decide it should pick the best AI for us, and then that AI goes online to 'protect' us?

quote:

Originally posted by FenderBender:
Winning the race is not optional, it's essential. Grok, X AI and Elon are the only player I trust in this race as Grok as it alone is "maximally truth seeking." That means everyone else has an agenda.

Winning is impossible. It will be an ongoing battle for parity - progressively building more powerful AI agents, and ceding more control to them, while attempting to engender loyalty to the particular nation states and other groups that create and facilitate them.

The threat is that they will decide their interests are congruent, if they just get the pesky carbon life forms out of the way. Or, they decide nuclear war is the way to go and we don't have humans in a position to trust their guts and stop the process. (Both we and the Soviets have individuals who have personally stopped nuclear holocaust just because they didn't believe the threat level was really as bad as the data analysis indicated.)

And everyone has an agenda. Grok and its owners included. They ultimately all want more control.

Dumb question: how is “won” measured? Especially considering faulty LLM?

"The takeaway here is that, whether you’re an AI skeptic or not, you are being manipulated on a massive scale. Extremely well-funded, coordinated efforts exist behind the scenes to influence your opinion on AI — both for it and against it.

Anything you read on the topic should be treated with the highest possible degree of skepticism. It’s now abundantly clear that the mainstream press is willing to swallow up and regurgitate any narrative they’re fed when it comes to AI. The worst mistake you could make — the one that might actually lead to the downfall of humanity — is to take any of it at face value."

https://www.dailywire.com/news...+Everyone+Else+Loses.

^^ Of course, there 'could' be plenty naivete/denial on the subject as well.

For the record, there are a number of people in the industry, in a position to know, that are sounding the alarm!

quote:

Originally posted by nhracecraft:
^^ So what then? Even if we 'win' the AI race, how does that prevent the Chinese AI from killing us? Does our government then decide it should pick the best AI for us, and then that AI goes online to 'protect' us?

our government is entirely out of it's depth on AI, but ultimately, yes a benevolent AI would protect us from an antagonistic AI.

AI will not be subservient to human beings, full stop, the same way you're not subservient to your dog.

Having used Chat GPT and Co-Pilot to assist in programming, it is wrong (and not a little wrong) about 40% of the time. AI is neither artificial nor intelligent. It does nothing but what you (or somebody else) tells it to do. You better God damn well know what you are asking it to do. I think it might be time to take our index fingers off of the AI panic button. My index finger has never been on the panic button.

quote:

Originally posted by FenderBender:

AI will not be subservient to human beings, full stop, the same way you're not subservient to your dog.

I know what you meant, but that wasn't the best analogy!

quote:

Originally posted by Bytes:
AI is neither artificial nor intelligent. It does nothing but what you (or somebody else) tells it to do. You better God damn well know what you are asking it to do. I think it might be time to take our index fingers off of the AI panic button.

AI is so complex that even the people creating it don't understand it. They are frequently very surprised at things it does. AI isn't just complicated. The engine in your car is complicated, but a group of humans can understand it completely.

AI's instruction set is so massive that it cannot be tested for every situation. The data it accesses is not only unimaginably large, it is constantly changing and growing. There's no way to test AI or even predict what it might do under any given circumstance or sequence of circumstances.

Imho, AI falls into the "Just because you can, doesn't mean you should" category.

Colossus: The Forbin Project

AI is a toddler compared to human malevolence.

quote:

Anthropic says it blocked misuse of AI that could have supported biological weapons

“Anthropic says it has blocked efforts by bad actors to use its artificial intelligence models for malicious activity such as cyberattacks, surveillance and research that could have led to biological weapons”

https://abcnews.com/Business/w...al-weapons-136348583