|
Transformers are not a monkey but it's going to be hilarious when the marketing furor convinces the judge there is a monkey in the box despite all expert testimony explaining it's just a big bag of tensors that calculate prompt = mush.
|
#
?
Mar 18, 2024 18:43
|
|
|
|
| # ? Jun 7, 2024 18:16 |
|
|
Bobcats posted:A lot of recent AI is falling into the scary as gently caress zone and my mom gifted me a cookbook for an air fryer and it’s blatantly authored by ChatGPT and the recipes are nonsensical. I hope they figure out dolphin language I'd love to throw shade on one of those assholes.
|
|
#
?
Mar 18, 2024 19:06
|
|
|
wake me up when LLM's have a solution to reversing entropy.
|
|
#
?
Mar 18, 2024 19:08
|
|
|
Roki B posted:wake me up when LLM's have a solution to reversing entropy. I asked it, it said "insufficient data for a meaningful answer." Piece of poo poo.
|
|
#
?
Mar 18, 2024 19:14
|
|
|
Roki B posted:wake me up when LLM's have a solution to reversing entropy. Easy! Reduce the size of the Universe.
|
|
#
?
Mar 18, 2024 19:15
|
|
|
Roki B posted:wake me up when LLM's have a solution to reversing entropy. redefine your system such that it’s externalized
|
|
#
?
Mar 18, 2024 19:18
|
|
|
Bobcats posted:A lot of recent AI is falling into the scary as gently caress zone and my mom gifted me a cookbook for an air fryer and it’s blatantly authored by ChatGPT and the recipes are nonsensical. If it's out of human text, what if the AI's start consuming each other's output and descend into generating ever more inaccurate garbage?
|
|
#
?
Mar 18, 2024 20:17
|
|
|
Clyde Radcliffe posted:If it's out of human text, what if the AI's start consuming each other's output and descend into generating ever more inaccurate garbage? as i understand it this is already happening
|
|
#
?
Mar 18, 2024 20:33
|
|
|
I came across this guy online who teaches ap world history as a career and on youtube. He then made an Ai of his own using selected sources. https://www.youtube.com/watch?v=wz_hEsxmEVs So Ai sucking up all the information online was a huge burst that pushed Ai futher. But like manufacturing where they revise a product using less materials, and improve the structures, Ai is doing the same. Ai doesn't need every single fan fiction to write a novel. Just toss every novel taught in an English classroom, and a few thrillers / romance novels, and you're good to go. fez_machine posted:Clarkesworld is still publishing. Ahh. Thank you mazzi Chart Czar fucked around with this message at 21:02 on Mar 18, 2024 |
|
#
?
Mar 18, 2024 20:54
|
|
|
mazzi Chart Czar posted:So Ai sucking up all the information online was a huge burst that pushed Ai futher. You can do that only if you build on top of a foundation model. And unless we want to just keep using the same foundation models ad vitam aeternam, we'll have to keep training new ones on presumably better generalized datasets. Losing our ability at cheaply-ish producing monstrously large corpuses to train those is a really big problem. We've tried building purpose-specific AIs for decades, and the recent breakthroughs are indicating that it's (so far) basically impossible to be good at anything non-trivial without having a strong general knowledge base to rest on. Aramis fucked around with this message at 21:12 on Mar 18, 2024 |
|
#
?
Mar 18, 2024 21:05
|
|
|
yep, those 2 years of gen ed and humanities credits at college were for a reason
|
|
#
?
Mar 18, 2024 21:06
|
|
|
zedprime posted:This isn't an AI unique problem. Ex. it is incredibly easy to make illegal settings in your cloud apps as a small business in privacy regulated industries. Cloud computing generally supports data handler regulations but also cars generally have seat belts and we see what the adherence is there when we let people set themselves up. I agree there's always loopholes or overlooked/malicious settings for cloud solutions, so it'll never be perfect. However, data privacy regulations generally are piss-poor, especially here in Australia, so it'd probably be safe to assume all generic Windows tech support can freely access the AI search/medical notes by default, until they're legally forced not to here. I'm a lot easier to sue than Microsoft or Apple directly. Co-Pilot is very tightly embedded into the OS structure, so I don't honestly know if there's a "Anything that even looks like medical files or records is sent to a different server with strong restrictions and barriers to access it, everything else goes to our generic server to for general tech support to access" special filter for Co-Pilot search data. If so, that's better than nothing, but I don't know that for sure. I'm very skeptical that MS or any other company would voluntarily do that for Australian customers, unless we made them. We don't have GDPR or other real data privacy regulations here: given the recent high profile data breaches, people are going to be even pissed off in the future. Don't MS, Apple & Android already copy and then follow their GDPR requirements for the Australian healthcare market to be on the safe side? In theory yes. According to conversations with school friends who worked IT support specifically for Australian healthcare providers, their actual answer was "We say we're GDPR compliant just in case any EU residents live here, but in practice, a lot of confidential medical data is accessible by default to the lowest IT workers here with no real safeguards or filters stopping them. If they access something they're not supposed to, they just get a reprimand & we desperately hope they don't leak the details of the medical file they saw".
|
|
#
?
Mar 19, 2024 02:07
|
|
|
Livo posted:I agree there's always loopholes or overlooked/malicious settings for cloud solutions, so it'll never be perfect. However, data privacy regulations generally are piss-poor, especially here in Australia, so it'd probably be safe to assume all generic Windows tech support can freely access the AI search/medical notes by default, until they're legally forced not to here. I'm a lot easier to sue than Microsoft or Apple directly. In other words if Copilot is sending data overseas illegally it's a hugely hilarious own goal (which, well, can't rule out) unless it's some really annoyingly novel case of trying to prove if tokenized data sent through a data cheese grater is still a protected data class. Because not an expert but the architecture of these things is you usually have a foundation model which MS is gonna set up in its beefier nodes, your computers own model which is the chopped up tokenized extra bits that can live in your windows install, and some compute in the middle, probably at your local node, with tendrils back to the foundation model and into your local model where it plugs beep boop tokens into the math equations to spit out some probably answer.
|
|
#
?
Mar 19, 2024 04:12
|
|
|
zedprime posted:I mean if Australia doesn't have the regulations many other countries do and Microsoft, Google etc. have their cloud infrastructure set up to meet the regulatory needs of the world. And not just GDPR where they only need to prove the cloud janitors in Abu Dhabi took a security course to be a data processor. There are many regulations around the world requiring locality of data and every cloud infrastructure set up has safeguards and big honking buttons to stop an administrator from sending military secrets to random nodes because of a fail over whoopsie. Copilot terms of service: by using this product you agree not to search or scan any documents that would get us in trouble.
|
|
#
?
Mar 19, 2024 05:58
|
|
|
Ive trained an AI to recommend which gasses are safe for me to release into my home
|
|
#
?
Mar 19, 2024 07:51
|
|
|
Lil Swamp Booger Baby posted:Ive trained an AI to recommend which gasses are safe for me to release into my home They can't be trusted to recommend gasses!! quote:A New Zealand supermarket experimenting with using AI to generate meal plans has seen its app produce some unusual dishes – recommending customers recipes for deadly chlorine gas, “poison bread sandwiches” and mosquito-repellent roast potatoes.
|
|
#
?
Mar 19, 2024 08:07
|
|
|
AI robot: "Okay so, you've asked me to jack your, uh, dick off."
|
|
#
?
Mar 19, 2024 08:54
|
|
|
Tarkus posted:Poke it in the eyes, easy peasy, defeated bear. ...
|
|
#
?
Mar 19, 2024 09:00
|
|
|
I can't believe Brian-san predicted this tech, so many years ago.
|
|
#
?
Mar 19, 2024 11:10
|
|
|
He's right you know
|
|
#
?
Mar 19, 2024 11:11
|
|
|
zedprime posted:This is a good example for a pet peeve of mine when we say a LLM isn't creative or can't apply anything to truly new situations. These statements are not not right but have an imprecision in language I kind of hate. Thanks for this, honestly food for thought.
|
|
#
?
Mar 19, 2024 11:29
|
|
|
My brother in-law tried to use AI to plan a drive to Guatemala and it insisted he go through Juarez, lol.
|
|
#
?
Mar 19, 2024 12:35
|
|
|
Das Boo posted:My brother in-law tried to use AI to plan a drive to Guatemala and it insisted he go through Juarez, lol. lmao
|
|
#
?
Mar 19, 2024 13:05
|
|
|
Bobcats posted:A lot of recent AI is falling into the scary as gently caress zone and my mom gifted me a cookbook for an air fryer and it’s blatantly authored by ChatGPT and the recipes are nonsensical. https://www.theguardian.com/technology/2023/sep/01/mushroom-pickers-urged-to-avoid-foraging-books-on-amazon-that-appear-to-be-written-by-ai
|
|
#
?
Mar 19, 2024 15:12
|
|
|
fish shaped crackers, fish shaped candies, fish shaped solid waste, fish shaped dirt, fish shaped ethyl benzene, pull and peel licorice, fish shaped volatile organic compounds and sediment shaped sediment (and) candy coated peanut butter pieces - shaped like fish.
|
|
#
?
Mar 19, 2024 15:17
|
|
|
Man Running AI-Powered Porn Site Horrified by What Users Are Asking Forquote:I hope they don’t actually want to see the things they’re typing in.
|
|
#
?
Mar 19, 2024 15:30
|
|
|
Lol
|
|
#
?
Mar 19, 2024 16:04
|
|
|
Currently, yes, OP. But every time anything new shows up, you can bet there will be a ton of idiots trying to cash in on it. This time most of the initial cash seems to have been breathless news articles about AI, followed by a bunch of financiers speculating on a new tech thing hoping they'll get a big payout. Some intelligent people (people I respect) have been warning for years about another type of AI that we don't yet have but which appears to be developing rapidly. When it hits, it's going to take a lot of people completely by surprise, because it will also be called AI but it will be radically different from the things we've become accustomed to when someone says AI. In the meantime, I'm using LLMs in a similar way to how pocket calculators were used when they became popular. I'm having to rewrite lots of my training because I can no longer assume a question is too challenging for someone to blithely enter it into a computer and have the right answer pop out. In this way, LLMs have eliminated the need for me to teach aspects of what I teach, similar to how calculators have eliminated the need for people to memorize multiplication tables in order to function in a society. I don't expect general education requirements to catch up with this new state of affairs for 50-200 years, though, because they still haven't caught up with pocket calculators. cruft fucked around with this message at 21:58 on Mar 19, 2024 |
|
#
?
Mar 19, 2024 21:56
|
|
|
It's only going to become more important to learn your times tables if you want to stand even half a chance against the linear algebra terminators.
|
|
#
?
Mar 19, 2024 22:16
|
|
|
cruft posted:Currently, yes, OP. poo poo, what's this other type of AI?
|
|
#
?
Mar 19, 2024 22:18
|
|
|
Rad-daddio posted:poo poo, what's this other type of AI? R&D teams didn't, like, close the briefcase with a smile and then go home when ChatGPT was released.
|
|
#
?
Mar 19, 2024 22:19
|
|
|
zedprime posted:It's only going to become more important to learn your times tables if you want to stand even half a chance against the linear algebra terminators. Once the drone wars start not much else will matter.
|
|
#
?
Mar 19, 2024 22:20
|
|
|
Clyde Radcliffe posted:Like right now we have scammers in call centres around the world trying to steal money from gullible people. What happens when they're replaced by an AI with a generated friendly local accent capable of responding more naturally than someone paid to follow a script? What happens when AI finds a way to bypass my "being a racist" filter? 
|
|
#
?
Mar 19, 2024 22:21
|
|
|
It is extremely cool how ai is ruining google as a search engine SEO was most of the nails in that coffin but AI is just speedrunning the burial, I am honestly impressed how lovely google is to use these days
|
|
#
?
Mar 19, 2024 22:24
|
|
|
cruft posted:R&D teams didn't, like, close the briefcase with a smile and then go home when ChatGPT was released. No, but the vast majority of them who haven't pivoted to Transformers are still on the same trajectory they were 10 years ago. Who's the outlier within that group is still an interesting question. For example, spiking networks are still as promising as ever... Just any day now... Aramis fucked around with this message at 22:27 on Mar 19, 2024 |
|
#
?
Mar 19, 2024 22:24
|
|
|
cruft posted:R&D teams didn't, like, close the briefcase with a smile and then go home when ChatGPT was released. no, they made 3 more versions of chat gpt and now they’re working on version 5 of the same thing
|
|
#
?
Mar 19, 2024 22:29
|
|
|
rotinaj posted:It is extremely cool how ai is ruining google as a search engine i'm having "to abuse" "air quotes" to get anything useful out of "the drat thing" at least the before:YYYY-MM-DD filter still seems to work, so you can at least get back to the low-background steel equivalent of content somewhat easily
|
|
#
?
Mar 19, 2024 22:38
|
|
|
cruft posted:R&D teams didn't, like, close the briefcase with a smile and then go home when ChatGPT was released. mm yes and which EA cult guy do you feel is the best candidate to give all our money to to ward off/appease Skynet this time
|
|
#
?
Mar 19, 2024 22:39
|
|
|
Come with me if you want to live.
|
|
#
?
Mar 19, 2024 22:39
|
|
|
|
| # ? Jun 7, 2024 18:16 |
|
|
At the risk of being made look like a fool by the robot overlord in my lifetime, every outfit who proclaims they are working on AGI for reals look to be doing the equivalent of the biologists electrifying big tanks of amino acids. Which is not unimportant work but not for the reasons stated by the guys writing the grants. Data modelling is one very specific avenue of data science oddly enough called data modelling. It's pretty cool sometimes but ultimately just one branch of one type of data management. gently caress modelling, tell me when we have a working librarian algorithm.
|
|
#
?
Mar 19, 2024 22:46
|
|