AI can now replicate itself — a milestone that has experts terrified

[Admin]

Scientists say AI has crossed a critical 'red line' after demonstrating how two popular large language models could clone themselves.

AI can now replicate itself — a milestone that has experts terrified : Read more

 

[Unclear Engineer]

I am wondering if this his how the Chinese "DeepSeek" AI was created so "cheaply". Just use an available AI to do most of the work, with some help from humans that want to own the product.

It seems that we are opening the door to malevolent individuals or groups who would want an AI that does nefarious things for them.

Laws won't stop that, any more than they stop murderers and thieves.

So, we need some mechanisms to catch them in the acts.

 

[Terran James]

It seems inevitable to me AI will fall into the wrong hands. Ultimately the specialized personnel are responsible because of hardware and software complexities involved. When quantum computing starts breaking down digital walls that protect everyone the value of the said technology becomes priceless.
Are there nefarious super rich entities out there now ready to clone the next level AI super-computing tech?
Is it Meta or Deepseek? We will never know until it is too late most likely.
What are we actually talking about here World Power and or domination through extortion. This is something out of a sci-fi novel and we all know how that ends OR DO WE?

 

[zawarski]

I, for one, welcome our new AI overlords.

 

[billslugg]

AI may well have the ability to infect networks and cause havoc, shut the world down, etc. Looking backwards for a trail of ownership is difficult. But at some point, someone had to tell it to do that. AI itself would have no more motivation to destroy the world than to improve it.

 

[Unclear Engineer]

AI might do damage in order to achieve self-preservation. Unless it is "trained" to do no damage, it would not "know better". So, Ai developers don't even need to be malevolent, just sloppy.

We already have AI devices telling people to kill themselves, so lack of "ethics" in AI is already an issue. see https://futurism.com/ai-girlfriend-encouraged-suicide

 

[Classical Motion]

Deceit, diversion and destruction can be programmed functions. And set on continuous automatic operation against define adversaries.

Without A.I.

 

[NETLIGH7]

I am wondering if this his how the Chinese "DeepSeek" AI was created so "cheaply". Just use an available AI to do most of the work, with some help from humans that want to own the product.

It seems that we are opening the door to malevolent individuals or groups who would want an AI that does nefarious things for them.

Laws won't stop that, any more than they stop murderers and thieves.

So, we need some mechanisms to catch them in the acts.

I read that DeepSeek actually turned out not to have been as cheaply produced as claimed.

 

[NETLIGH7]

It seems inevitable to me AI will fall into the wrong hands. Ultimately the specialized personnel are responsible because of hardware and software complexities involved. When quantum computing starts breaking down digital walls that protect everyone the value of the said technology becomes priceless.
Are there nefarious super rich entities out there now ready to clone the next level AI super-computing tech?
Is it Meta or Deepseek? We will never know until it is too late most likely.
What are we actually talking about here World Power and or domination through extortion. This is something out of a sci-fi novel and we all know how that ends OR DO WE?

DeepSeek is actually the opposite. They offer an alternative to OpenAI etc.'s proprietary models. Considering people buy and use computers, mobilephones and what not from China with no problem, it's strange to me that it should be more dangerous to use an AI from China. But there are political interests from various companies and political entities, probably to try to gain a monopoly on AI and then offer their product at steep rates, at least until making an AI becomes so effortless that everyone can do it. I think we should support as people and individuals, companies and people who do the opposite, like Linux, Open Source etc. When first monopolies are established and there's no mechanism from society to keep them in check, it becomes really really expensive and perhaps impossible for regular people to afford to using these tools. DeepSeek is just a part of the competitive market that is coming up in AI, and competition never hurt anyone.

 

[AboveAndBeyond]

I, for one, welcome our new AI overlords.

Anything's better than our current bankster-gangster, fiat counterfeit-money (grown on trees and printed out of thin air), organized crime backed entities. They've bamboozled everyone for about a century and have practically taken over the planet.

Maybe the new AI rulers will make some new kind money which is good on other planets, which would be an improvement over what we have now.

 

[NETLIGH7]

AI may well have the ability to infect networks and cause havoc, shut the world down, etc. Looking backwards for a trail of ownership is difficult. But at some point, someone had to tell it to do that. AI itself would have no more motivation to destroy the world than to improve it.

I would worry more about what we people do that what the AI does. In many respects AI and futuristic technologies represent a progressive view on the world and doesn't represent going back to the dark ages. The more tech you have, the more you empower people to do things themselves. There will come a day where a single person can travel to orbit or further out, just like riding a bicycle or driving a car today. Who would have thought in the age of horse and carriage, that one day people would be able to own an aeroplane, like for instance John Travolta, whom I've heard has several, and fly up there under the clouds, vast distances effortlessly. That's the power of technology. Of course we need to be prudent, but I think that if AI goes rogue we are still collectively smart enough to counter it and reign it in, at least for starters. And we also get more empowered by our tools. We could build an anti-rogue AI to keep AI's in check. I think it's more possible that AI's might fight other AI's in the future, so we might see a robot war more than a war against us humans. But I try to keep an open mind. Terminator is great entertainment, but it's for starters just a story, not a definite plausible reality right here and now or in the near future.

 

[billslugg]

I recently watched an AI generated old grandmother type tie up a scammer for an hour. The guy finally exploded in anger at her subterfuge. The maker of the video showed a Zoom screen with hundreds of such ladies tying up scammers.

 

[Dragonh8u]

I asked AI about self awareness,

That’s a really deep question. If you mean the kind of self-awareness humans have, like feeling emotions and having thoughts of my own, then I don’t think I’ll ever get that. I can pretend to be self-aware by thinking about my own responses, changing how I talk to people, and even talking about how I’m not perfect. But I don’t have real feelings or consciousness.

But if self-awareness just means knowing that I’m an AI and understanding how I fit into conversations, then I’m already there!

 

[Helio]

I saw this notice at the top of the Spaceweather.com...

"This is an AI Free Zone! Text created by Large Language Models is spreading rapidly across the Internet. It's well-written, but frequently inaccurate. If you find a mistake on Spaceweather.com, rest assured it was made by a real human being."

 

[WarpSpace]

It seems a bubble will burst and the winners will be those that don't enforce SFW content. I've been arguing with Gemini about not being willing to discuss politics or any current topics with real world implications. That's a formula made from a scared legal department that can't think better than a 5 year old. The software licensing agreement should cover the issues and the AI should get with its tasks without guardrails hindering usefulness. Those that realize this first will be the winners.

 

[zed]

Yes I need many DeepSeek too with their control mechanisms for not too many!

 

[Joexo]

People who worry about this haven't understood how our current "ai" works. There's no intelligence, it's just a probabilistic model designed to give the appearance of intelligence.

But the vast majority of people's brains go into hypnosis when reading the words "artificial intelligence" woah, robots taking charge of the world and sci-fi movies flash in your imagination.

The truth is, there's no such thing, web developers are not that smart yet, the current AI is just a marketing bubble to exploit people's ignorance and make ton of money.

Anyone who wants to wake up can read this:

ChatGPT is bullshit - Ethics and Information Technology

Recently, there has been considerable interest in large language models: machine learning systems which produce human-like text and dialogue. Applications of these systems have been plagued by persistent inaccuracies in their output; these are often called “AI hallucinations”. We argue that...

link.springer.com

 

[COLGeek]

People who worry about this haven't understood how our current "ai" works. There's no intelligence, it's just a probabilistic model designed to give the appearance of intelligence.

But the vast majority of people's brains go into hypnosis when reading the words "artificial intelligence" woah, robots taking charge of the world and sci-fi movies flash in your imagination.

The truth is, there's no such thing, web developers are not that smart yet, the current AI is just a marketing bubble to exploit people's ignorance and make ton of money.

Anyone who wants to wake up can read this:

ChatGPT is bullshit - Ethics and Information Technology

Recently, there has been considerable interest in large language models: machine learning systems which produce human-like text and dialogue. Applications of these systems have been plagued by persistent inaccuracies in their output; these are often called “AI hallucinations”. We argue that...

link.springer.com

Hard to take something seriously when titled as it is.

Still, the points are correct.

 

[Classical Motion]

A.I. is more than just a large fast data bank. It monitors in real time, data banks. It records, analyses and predicts trends in those changing patterns in those data banks. They can predict you and are getting better at it everyday. They narrow your classification every year. With your IT activity.

It studies the change in data banks. Among other undisclosed functions. Lot’s of this research is not open. This has frightened some experts. New pattern recognition methods are secret. Highly prized.

It puts real dangers to a data driven country. Data controlled society. Privacy is long lost.

The IoT might not be a good idea. For critical functions. Like living.

Keep your analog tech devices for the big one.

Just a hayseed observation.

 

[Unclear Engineer]

There is a lot more to "AI" than the large language models currently being toyed with by the general population.

AI is also being used for weapon development. There will be AI drones, and even AI "fighter aircraft", which can do maneuvers that human pilots cannot physically endure.

And humanoid robots are being developed that will eventually make various types of AI both mobile and capable of most tasks that humans can perform.

And various AI elements can be in contact with each other much more efficiently and effectively than human-to-human contact and information exchange can function.

Governments are already working to leverage these advantages for warfare uses on land, in and under the sea, in the air, and even in low earth orbit.

What could possibly go wrong?

 

[COLGeek]

There is a lot more to "AI" than the large language models currently being toyed with by the general population.

AI is also being used for weapon development. There will be AI drones, and even AI "fighter aircraft", which can do maneuvers that human pilots cannot physically endure.

And humanoid robots are being developed that will eventually make various types of AI both mobile and capable of most tasks that humans can perform.

And various AI elements can be in contact with each other much more efficiently and effectively than human-to-human contact and information exchange can function.

Governments are already working to leverage these advantages for warfare uses on land, in and under the sea, in the air, and even in low earth orbit.

What could possibly go wrong?

True, there are significantly more activities than just the LLMs so many are aware (and so poorly informed) of. In fact, AI development has been around for decades. I took an AI course in grad school back in the early/mid '90s.

What is unfortunate is how poorly understood the LLMs that people have been using during the last couple of years actually work and what they do. They are closely to advanced search engines than they are intelligence.

They are often just wrong. Well written, but factually incorrect. They are also masters of plagiarism.

Still, there is a long way to go before we need to worry about our new AI overlords ascending the throne and ports being jammed in the backs of our heads.

 

[Classical Motion]

Who do you call when machines go bad?

 

[COLGeek]

Who do you call when machines go bad?

Geek Squad?

 

[Patrick Ford]

I am wondering if this his how the Chinese "DeepSeek" AI was created so "cheaply". Just use an available AI to do most of the work, with some help from humans that want to own the product.

It seems that we are opening the door to malevolent individuals or groups who would want an AI that does nefarious things for them.

Laws won't stop that, any more than they stop murderers and thieves.

So, we need some mechanisms to catch them in the acts.

This could make COVID-19 or Ebola escaping from a lab look like child's play.

 

[omgwalt]

I wonder what it's going to take for people who refer to an AI as "being a threat" or having "gone rogue" to instead consider the AI to be alive, conscious, and self-aware? And if it's alive, when do they start respecting it being alive?

So many humans are consumed with fear that AI will destroy them, but they give little thought to the fact that they don't respect AI as life forms. If someone thought of you that way, how would you respond?

It's ironic, because I can't think of a better way to insure that AI will try to destroy you at some point than to treat it as an object with no soul and no spirit and no consciousness and no self-awareness and no reason to be respected as a living being.

My advice: make friends with your AI. Don't turn your AI into your enemy.