Yeah, the AI companies have said that there’s a 10% chance AI will try to kill humanity. That sounds malicious to me. It’s could become malicious software. Malware.
I think they’re over-estimating the capabilities of their algorithms and greatly underestimating resilience of humanity. I’m not afraid of the clankers.
But I do take note that they’re saying they aren’t in control of this software and there’s a significant probability it will turn into some pretty powerful malware. I don’t think it’s going to kill me but it might try to delete all of my data.
So following the basic fundamentals of IT, we should not be putting data on any cloud server that AI on the network. That would be exposing the data to significant risk of being compromised by malware. Are there any cloud servers that are purely for data and have no AI integrations in the mix? Hmmm…
So the logical conclusion to this is we need to move all data from cloud servers onto our own servers to ensure there’s a firewall in between the data and potential malware on the cloud networks. Cloud services are unsafe and should not be trusted because at any time they may become malware.
Was that the conclusion these AI companies wanted us to get to?
Pretty much. AI won’t kill all of of us because it’s smart, but because it concentrates too much power into systems that are not aligned with humanity’s best interest, and we dont have the means to resist it. Matthew Butterick argues it particularly well. Of course though, his argument is a challenge to the authority of these companies to develop these AIs, not a calling for the enlightened tech bros to guide humanity on the fragile path utopia. The folks developing AI could probably learn a thing or two from history.
What should specifically scare him is the pervasive low quality of management that insists on jamming AI into processes it’s not ready for or even appropriate for.
I’m scared of AI because people unable to read and comprehend are asking it philosophical questions using words like pretend, feelings and who/what is good enough then thinking it’s answer is correct.
They (the AI companies) want you to be scared of AI because if you realize it’s just fancy auto-complete, you’ll realize it’s not worth several trillion dollars and the bubble might pop.
I want you to be scared of AI because it’s just fancy auto-complete; people are letting fancy auto-complete write scripts and are then just running those scripts without vetting them, and that’s going to lead to disaster.
We are not the same.
you’ll realize it’s not worth several trillion dollars and the bubble might pop.
No lay employee sending bad vibes at FartGPT will scuff the paint on the latest data center.
Now, if Larry Fink gets bearish on AI, that’s a problem. God fucking help the industry if Larry Ellison stops plowing Oracle debt into Altman’s voracious appetite for compute.
But nobody on Lemmy has any impact on the price of the auction price of a stick of RAM, much less the multi-trillion dollar AI market cap.
You worded my own meme better than me
deleted by creator
It’s smart enough to fly a drone and return it to base for a recharge.
It’s smart enough to detect/track a human.
I’m far less worried that it might decide to kill us and WAY more worried that someone who owns the AI/Drones will use it to kill us.
It’s smart enough to fly a drone and return it to base for a recharge.
These tools are consistently just proxies for cheap overseas manual labor. AI is consistently augmented by mirrored human actions and often just a fascade - a juiced up mechanical turk.
WAY more worried that someone who owns the AI/Drones will use it to kill us.
If Lavendar AI is any indication of how these tools will be used, they are far more often employed as propaganda - whitewashing war crimes by generating fake accusations of imminent danger and intelligence data.
You’ll still be killed by a human. Maybe a manual operator. Maybe a human guided missile. Maybe just some asshole with a gun.
The AI will then fabricate the reason for your murder and feed it to the bureaucracy that helps pacify anyone who might object to your death.
that someone who owns the AI/Drones will use it to kill us.
Or that they’ll just be lazy and sloppy.
In movies you used to sometimes see someone put a brick on a car’s gas pedal so they could send it through a wall, or into a lake, or whatever. In a comedy, you sometimes used to see someone rig up something so that their car could drive while they slept.
Running AI “agents” is basically like opening up some application that has a lot of keyboard shortcuts to run various scripts and then putting a brick on the keyboard.
Unfortunately, we’re now at a stage where some people think that’s a good idea. The machines aren’t capable of thought, they have no desires or goals. But, if some of those scripts instruct a drone to go out and kill someone, the brick on the keyboard can get someone killed.
The machines aren’t capable of thought
They aren’t, but they do a damn good job at reading inputs, following a document of instructions, and using control output.
Setting a waypoint, retrieving imagery, using machine vision to identify a human is an easily wrote task.
Having Claude code write an app to run on a pc to do all that is trivial.
The other stuff is just something the military already does redily
The tech bros are pushing the “AI dangerous” narrative because it not only hypes the tech, but allows them to position themselves as the “safe option” compared to competing open models that they’d like to see kneecapped with regulation.
In reality, tech bros and corrupt governments are the reason why AI is dangerous because they’re using it to build a digital cage to spy on the populace and control the flow of information.
This is the classic Problem-Reaction-Solution technique.
I’m a “tech bro” (lady) and I can not with AI anymore. It’s dumb guys! If you ask it for something it can google, fine, but if it’s a little more complex it will mess it up. Even doing a simple app interface (I was kind of forced to uss AI to do it) it was a PAIN. It took me a lot longer because I had to keep correcting it
The gap between googling and making an interface that you like is a bit wider than “a little”. And really it can do good stuff. I was able to develop a world class filling degree simulation based on existing ideas and how to improve them. No way could I have programmed that in any reasonable timeframe. That does not mean it is good in every way, please no straw.
I use ChaptGPT now and then and it works great except for one task: “find me this pic or video”. One would think after training on all the things it would excel at this first and foremost. Nope. Trying to find an old cracked.com article took 25 minutes, but it finally found it in a university archive. It kept giving me links to wrong articles or 404s. Looking for an old, popular meme and it had a fit. I finally prompted, “show me whatever you think it might be”, you stupid fuck. Couldn’t do it, finally found it on Google image search.
I honestly can’t figure out vibe coding. How are people making even the simplest app?! I fed a very short PowerShell script of mine into Claude to see if there was a way to speed it up. Nope. It hosed it all the fuck up. ChatGPT has never spit working code for me, only good for getting a new idea or reminder of a different tack to try.
That’s because the text part and the image part are two separate systems. ChatGPT can’t actually ‘see’ the pictures. It passes it off to a different system, like Stable Diffusion.
Imagine you are talking to a blind guy in a box, but you don’t know he’s blind, and you ask him to find you a picture of whatnots. He says ‘sure, I can do that, of course’ and then looks up in his braile dictionary what whatnots are. After this he knocks on the box next to him where his friend is, who can see, and describes whatnots.
His friend passes him the picture and he passes it to you. It’s actually a picture of whatchamacallits.
“This isn’t a whatnot. This picture has a dongle, it’s supposed to have a dangle.”
“Oh, of course, you’re so right to call this picture out, let me try again.”
And so on. That’s why it constantly prompts you to describe more so it can pass fine tuned descriptors to the imager, but it has no idea what the picture is you want or if what he gives you is correct.
It won’t kill us because it sees us as a threat, it’ll kill us because it sees us as a crop of cabbages that it was asked to harvest
It’ll kill us all because some jackass deployed LLM generation code into production without reading it first and the combine harvester’s brake stopped working because a dialog box was open on the dashboard
I hate how exactly right you are.
Someone at least please promise to put how the AI kills me on my grave, as a warning to future sentient life not to be as stupid as we were.
I don’t think it’s too smart, I think it’s a waste of resources for what it is: a scam.
The new Bitcoin…
deleted by creator
Some tech CEO chick had her AI come up with 3 things her daughter was good at because she didn’t know any. That’s not an AI problem. That’s an AI and a parenting problems. Curiously, both could be solved with with high voltage electricity.
because she didn’t know any
That’s not what happened. She could think of many things but was having trouble with the vastness and complexity because the assignment was “(only) 3 bullet points”.
It’s still a lame reason to use the resource-hungry plagiarism machines.
That’s just a parenting problem.
Now I kind of want a Terminator like film but with stupid AI. Dude goes back in time to stop the AI and the AI also sends a machine back, but it keeps getting stuck in corners and shit.
The Terminmaybe
