I know there’s a significant anti-AI presence here. I’m not promoting it. I found it interesting to read how it was used, it’s strengths and limitations.
This isn’t an LLM, which is the thing most people are hating against. Pattern recognition is what AI is actually good at.
This is what it’s supposed to be used for. Things like this are where LLMs are a benefit to society. The ability to “intelligently” process incredible amounts of information fast lends itself to things like this. Or helping Air Traffic controllers do their often hectic jobs, or sifting though the trillions of gigabytes of data that come in from space telescopes looking for anomalies, etc… etc… etc…
THAT would make me excited about AI. More…of…this…
But instead they use it make people even more lazy, and make corporations able to make billions more by firing real humans.
Yeah, but have you considered that this kind of technology can’t sustain the delusion idea that you could replace all workers with a chatbot?
Note this isn’t an LLM.
I was about to say. Couldn’t imagine an LLM would be great at stuff like this.
Yeah, this is right up machine vision alley, and is a pretty straightforward and good thing in general. It lets computers look through haystacks that humans just would take forever to get through. A thoroughly attentative human may do as well, if not better, given enough time, but it takes a lot of time and a human will burn out before reviewing that much imaging.
Dear God please do not let any of the current iterations or any of the next or even the ones after that of an llm anywhere near an air traffic controller tower. Fucking swear to God we will have planes smashing into each other when they start hallucinating their asses.
Oh, believe me. I agree. But this is what they SHOULD be working towards improving LLMs for, NOT better ways to steal art.
LLMs are not decision makers. Use a different model type if that’s what you need. LLM’s best aspect is their language interpretation, it doesn’t do much else
The AI picked through the pictures taken by the drone pilots pixel by pixel, looking for anything that might look out of place on the mountainside. The software identified dozens of potential anomalies from a large number of photographs in a matter of hours.
The selection, however, still needed to be whittled down with some human expertise.
“The software could react to different things, like a piece of plastic garbage or an unusually coloured rock,” says Isola. “It can even hallucinate some things. So, we still had to narrow it down further by taking into consideration the path that Ivaldo, as a very skillful climber, might have used.”
Interesting process. “AI” as a term gets so overused, but in this instance I think they’re really talking about image neural net processing.
This other one mentioned sounds like just image processing:
Other software that searches for unusually coloured pixels in natural landscapes – developed by the Lake District Search and Mountain Rescue Association in the UK – has been used to locate the body of a missing hillwalker in Glen Etive in the Scottish Highlands in 2023.
Or is it ML, not AI?
The key is to keep training the machine learning systems that power these algorithms to improve their accuracy in different types of terrain and conditions, says Tomasz Niedzielski, an expert in geoinformatics at the University of Wrocław and leader of the team that developed the SARUAV software.
Overall interesting process but could be a lot more specific about the technology.
ML is also AI. Obviously would be one of those instead of the LLM stuff, and generally the LLM stuff is the more controversial.
Not really. We have had access to ml for a while and google used ml to blur license plates in streetview for a long time now. It’s just pattern recognition with reinforcement.
Neural learning has been around since 2004-ish.
Yes. And I took an “AI” class as part of computer science curriculum back around 2000, including implementing “AI” stuff in lisp. We’ve been talking about AI for decades and ML for machine vision was always under that umbrella from the time out started becoming viable.
LLM is the recent popular subset of AI, not all of AI.
AI is a very big umbrella that encompasses a lot of methods computers can use to simulate decision making. Machine learning is just one of those. So are behavior trees for video game bots.
ML isn’t AI, but a lot of people (companies trying to make money) have started branding it as AI.
I mean, prior to the whole LLM craze ML was pretty much ‘the’ AI thing.
Note that AI got so messy they started saying the thing people imagined is ‘AGI’, so I don’t know if you are thinking that, but in that picky scenario there is no such thing as ‘AI’ on the market with that definition.
deleted by creator
I mean, prior to the whole LLM craze ML was pretty much ‘the’ AI thing.
In pop culture, perhaps.
I would say more the opposite, in pop culture, LLM == AI. In the technical world, both in university and industry, AI has covered a lot of areas, and machine vision based on ML was absolutely under the category of AI. If you said you used pytorch to train an AI model no one in the industry or academia would have batted an eye.
No, not at all. AI has always meant more than just ML or any subset. Crack Russell and Norvig’s book. Yes, it was written from an intelligent agent perspective, but absolutely not limited to any form of ML.
You go to the major AI conferences—AAAI or IJCAI, perhaps, and you will find a wide assortment of studies that have nothing to do with machine learning. Logic foundations of AI? Knowledge representation/reasoning? Decision-theoretic planning? NLP? AI has a ton of relevant subfields, and entire conferences that have nothing to do with ML.
Hell, for years at major conferences we’ve seen live contests in areas like SAT solving and game playing (based on planning not matchine learning).
I’ve spent my entire career studying artificial intelligence, and very little of it would be classified as machine learning.
Ok, so you weren’t countering that ML wasn’t AI, just contending that ML wasn’t “the” AI in the AI field, that’s fine. Keep in mind this thread kicked off from an assertion that ML wasn’t AI. Fine, ML wasn’t all of AI, but definitely was AI and in the popular understanding, it was pretty much “the” AI in the same way LLM is “the” AI now.
This reminds me of that old SCP where some photo of a mountain had like 4 pixels of the “anomaly” in it, which would trigger if you noticed the 4 pixels. SCP-096 I think. Haven’t been on that site in ages, lol.
Yup. And there’s a pretty fantastic short movie about it on youtube called 096.
So stuff that already existed more than ten years ago.
Basically, yes. Just more sensitive.
Sorry friend. Lemmy will always downvote anything that potentially paints AI in anything but a negative light, regardless of the context
Yeah, I expected it. I guess I don’t see things in black and white anymore.
Basically, yes. Just more sensitive.
Do you really…
I have no fucking clue what you’re trying to say
Ypu didnt expect it and still see things in black and white. The fact you werent able to process the previous post proves my point.
You’re quoting a different comment than this comment chain and you’re replying to me as if I said the black and white comment, which I didn’t. I don’t know what point you think you’re making, but you couldn’t be more ineffective if you tried.
