I was working on this 7 years ago or so. Either no one was interested or I just genuinely suck at sales and marketing so I gave up on it. Might have a buyer now so I'm updating the code base
I was AI to price fix ALL groceries at a specific low profit level (2% tops) at the grocery store NEAREST to me, keep the jobs at the store sustainable, balanced, and the store’s property & business taxes to be set by the city by AI, so their tax level is sustainable for maintenance on sewers and roads.
I want AI to catch and punish corruption. I want corruption to be excluded from pardon (effectively treated like declaring bankruptcy) wherein the corrupt no longer able to commit corruption.
we desperately need to stop lumping all the weird shit under the "AI" label. Chat gippity is an LLM. Image generation is stable diffusion. machine learning and neural networks have been around for decades and have genuinely changed the world (sometimes for the better). they all apparently get called "AI", and a lot of "AI" fans transfer credit from ML/NN and give it to LLM/SD, stolen valor style.
If you're giving a computer some appearance of intelligence, it's AI. Videogame AI doesn't necessarily involve much more than some pathfinding algorithms and a set of weights for aggro, but nobody has a problem calling that AI. AI is the umbrella term. LLM, SD, ANNs, all types of AI. Hell, a nested set of IFs can be AI.
It's really not worth getting a whole rustled set of jimmies over.
not realizing that the average person doesn't grasp the technical nuance of neated terms is literally part of the problem, so thank you for highlighting the discrepancy between "a bunch of boolean strings is technically AI" and the average redditor genuinely perceiving the usage of the word "AI" to literally meaning thinking machines with souls and feelings who can actually talk to them, and aren't just churning out high-rank data nodes.
I just want the OS1 from the film HER. I just want someone to talk to. Sadly, we'll never get to that point. All AI being built is strictly to spy and gather information to exploit people for profit to make companies rich. Not for the human experience. Not to be helpful, or to improve our lives in any meaningful way. Just spy, profit, manipulation, and control of the masses.
The closest thing we have to "HER" today, is Nuro-Sama. An at-home project by a streamer that isn't made to be a personal thief for profit. Just a human-ish AI with a semi believable personality.
I asked chat GPT to go to a particular website and find me five meals that had to meet certain criteria and have overlapping ingredients. Then I asked it to make me a grocery list. It did really well on the first part, but it left out about half of the needed ingredients for the recipes, I think it only gave me the overlapping ingredients for some reason. I'm going to keep playing with the prompt but I'm almost there.
AI/robotics is for hazardous work in the mines, repetitive motion in the factories, maybe product inspections, shipping route calculations, NOT art, music or creative writing. Those things are what the HUMAN MIND is for.
Literally you are. All the artwork of every field was scraped and put it into the gen ai programs. None of that belongs to the techbros or the equally exploitative userbase.
I just use AI to rewrite what I already wrote. Because my autistic ass writes something and it's filled with parenthesis and extra thoughts and overly detailed information and run on sentences so that neurotypical coworkers actually read the entire thing instead of giving up a paragraph in.
Honestly sounds like something AI would write. Literally no one needs AI to anticipate what they need to restock, and they certainly aren't going to drive to 4 different locations for the best price on several different items. If there are people out there that think this is a better use for AI (and the resulting waste of resources), humans have become a race of idiots.
I kind of use it for this very reason. I ask it to limit itself to three stores near me and ask which has the best price on size/version of product X. sometimes it doesn't always work and I have to check manually, but.. it has cut the time in half of doing the entire grocery list.
In reality, it just wouldn't do what OP asked for. It probably would even seem like it works pretty well... except... it would just give you a list of sponsored products. Because tech companies don't build products anymore. They build facades that extort money from other companies.
But how do you expect it to anticipate that without studying your habits? People are reluctant to feed more data to the machine but this would require a LOT of data. On some level, you'd need to be okay with a machine observing you.
Camera in the fridge, use ML to identify what products are there. You can set up stock limits (e.g., "12 eggs", "2 bottles of milk", etc) and when requested, it produces a list that fills up to meet the limit (2 eggs left = 10 more needed, 1 bottle of milk left = 1 more needed, etc). It doesn't have to be complicated or manifest 1984.
Yea but that is exactly my point--you'll need to be okay with a machine observing you and creating a profile on you. also a single camera would never capture everything in a full fridge. It would have no idea how many eggs are left in a carton. How much sour cream is left in the tub. Etc. It can infer, by how many times its been removed and compare to historical data, but never truly know.
Instead they’re using AI for ‘dynamic price tags’ that are connected to Wi-Fi, have a camera for your face, and will profile you on the spot for the price. It will also raise prices to scalp to the it’s best ability.
/gallery/i-made-thing-4vQOJ Here's the post that died in usersub from it. It had vision AI and could already correctly id items in your fridge. The theory is that farm-direct would always be the cheapest option if available. I got a restaurant food distributor interested in it two weeks ago, so I'm updating the code base and working on ERP integration. I'll just keep tinkering with it so long as I stay unemployed.
No, but, collaborators do, and they turn into partners. Think of it like commissioned art. You know what you want to see, just can't put it on paper. Artist can put it on paper, just needs a blueprint.
Also money. You need either money, connections or skill to get such a partnership. A mere idea won't get you anywhere unless you patent it, in which case it falls under skill.
That's not actually a job for "AI", that's just the perfect job for regular old-fashioned ADP (Automatic Data Processing. What we had before fuckos and goobers came up with "IT".) The reason we don't have it is because capital owns the world.
Nah, the reason we don't have it is because there's not a good way to automatically track how many eggs you have left and the fact that there's only a half a glass of milk in the jug.
And also because there isn't a good way to track what's available where and how much it costs. Most stores around me don't have all their items listed on their websites. And even if they do, they don't specifically list what's available at the one closest to me. I don't know how AI would fix that, but AI can certainly look at a picture of a fridge and list what's in it.
Funny you say that. I can't put in a collect order without selecting the shop I'll be picking up in, and not just for the obvious reason but also because prices differ regionally, and because availability differs between shops.
Traditional automated data processes all rely on strictly defined data formats and sources. The thing with AI isn't that it does the same thing traditional methods do, it specifically has an edge on not requiring a specific format. It's absolutely not something that was doable with "ADP".
Yeah, kind of, but in this case I meant "not doable" as in not actually feasible as a product that people can use. It's specific to the person, or at least the neighborhood, it takes a lot of manual work to configure it for each shop you want to include, it likely needs a lot of maintenance, and each use-case has to be separately defined. With AI, you can skip all that. Today you want to know where to get baking products, tomorrow you're interested in gardening tools, and next week you need
to replace your jeans? an AI system designed for that can handle it, no problem. But with traditional methods? Each shop has to be painstakingly added and processing methods implemented. Not to mention any kind of intelligent system that tracks what you need(this doesn't exist yet, but it is feasible with AI agents).
Now to be clear, I don't think you should use AI for this because of various legal and ethical reasons, and I also don't believe that the current LLM/LRM approaches are not ideal for this kind of daily task assistance work. But the fact is, they do make it feasible to implement a system that works with enough accuracy to actually be usable, as opposed to traditional methods that simply aren't doable within a moderate amount of effort from the user and are absolutely not scalable.
Yes, the process is Automatic Data Processing, but the input output device I want to initiate that ADP is a simple assistant, to whom I can say, "Check my pantry, and make a shopping list. Prioritize least number of shops with max number of sales, coupons, and deals."
So many things are marketed as AI. I've spent 20 years in tech and it's just silly to watch. Especially those naive regarding what it can and can't do, or believing it understands just because it can regurgitate things it's taught.
Yep. Not a single other engineer ive talked to has anything good to say about it. We all realize its an overhyped predictive text run by statistics that cant really do anything well except help with writing. Every non technical person ive spoke to thinks its magic sentient machine that actually knows things.
We're fucked because no one can be bothered to even watch a 30m video on youtube explaining how it works.
The reason we don't have it is because the AI the post describes is a data-privacy nightmare.
"Yes, please, I would like for you to keep track of everything I purchase and own, my eating habits, geolocation, and purchase habits. I'd also like for you to decide and shove 'coupons' based off of that information. Also also, please know literally every store, deal, price, and location in the region, and decide for me which will better fulfill my needs."
Honestly like this already happens Tho. I was able to link my kroger savings account to instacart to get the same savings benefits. I enter my phone number wherever I shop these days. I like it because with modern digital tracking, you can pretty much find and verify every single transaction. If there's a financial problem or a product problem it's very easy to track.
And why is privacy so vital? Because capitalists will exploit every possible perceived weakness in order to extract more money, an imaginary resource, from everyone else on the planet with no regard to the real harm they do in the process.
Matter of time. Your old Nokia 3310 had orders of magnitude more computing power than what we went to the moon with. Your current PC blows turn off the century supercomputers out of the water.
In 10 years, a current computer will be able to run a pretrained LLM of today's quality, no problem.
It's comical how many people will, in one breath, hate that AI is being used to gather personal information, and then in the next wish that they had an AI that knows not only their own information, but also all the information for the surrounding few towns so they can get deals.
Not to mention, the type of AI that does that has nothing to do with generative art. It's completely different people making completely different products, but people just see the word "AI" and think its all one thing.
The underlying software for route planning and optimization and all that probably wouldn't be done inside tge LLM, however searching fliers and the like is absolutely within the realm of possibility for one, as is using one as an 'agent' to perform all the other clerical work that would go into something like this.
LLMs are extremely flexible due to the fact that 'language' can be any means of information encoding, afyer all they're predictive algorithms, there's no real comprehension.
If it were an AI it would be the same. The same a John Doe searches in his citys shops informations for the best deal and paints a picture in the evening. Otherwise its not an artifical intelligence but an artifical stupid.
Intelligence: the ability to acquire and apply knowledge and skills. AI is, in fact, definitionally correct. It's an intelligent tool (one trained to solve specific sets of tasks. Be it predict text, how proteins fold, or draw tits)
But people seem to have this weird habit of thinking "if intelligent, then needs to be intelligent in all things". As if nowadays we don't have plenty of examples of people who're geniuses in one thing and morons in another.
On top of this, AI is still in its infancy, in all current variants.
People are basically asking "Why can't my baby walk yet, what good is a kid who can't walk?" when it just barely made it to toddler status. We just barely taught the thing not to shit itself, needs time before it becomes capable of true greatness.
The fact that grifters are busy making money off of its current variant is a testament to the giant it is going to become as people work out the kinks behind the scenes.
"Whenever people ask me why I (contrary to widespread myth) actually like AI, and think that AI (though not GenAI) may ultimately be of great benefit to humanity, I invariably point to the advances in science and technology we might make if we could combine the causal reasoning abilities of our best scientists with the sheer compute power of modern digital computers."
I would also consider how many jobs are being lost to AI right now. I am constantly triggered by seeing Republicans talk about "illegals stealing our jobs" while literally AI is openly doing it for everyone to see. I'm probably going to be retraining for a trade because of that, and it's not making me a happy camper ;.;
I technically agree that would be awesome. But considering current AI is nowhere even remotely up to that point it feels a bit weird to defend current AI with this pipedream.
The article is not defending current AI — it's about why LLMs are _not_ going to be able to do this. The above quote is not a good summary of the article. Another quote, from the article: "But anybody who thinks LLMs are a direct route to the sort AGI that could fundamentally transform society for the good is kidding themselves."
It is, look up AI on protein folding, chatbots are just the lowest form of current AI, wasteful unsustainable and innacurate, the fact that they are the face of ai in the general public is just a marketing stunt for openai to keep burning billions, yay capitalism
So technically you agree it would be awesome but you believe it's a pipe dream? So are you saying that you believe it will never reach a point where it could be a benefit to humanity? And if so you must then believe no human will ever be able to make AI functional to the point of achieving this goal. If that is what you're saying you must believe you know more than all the people working on advancing AI that do believe they can accomplish this goal. Do you know more?
So are you saying the pipedream you're referring to is not that AI could, "ultimately be of great benefit to humanity"? It sounded like you said that was a pipe dream to me. What was the pipedream you're referring to if I made an assumption? It seemed like that's what you said to me and was therefore not an assumption. Maybe you can explain what you meant using words from the context in which your reply was made?
He said "It's weird to defend current AI with this pipedream" The pipedream, as defined in the statement, that AI could one day be of great benefit to humanity. Pipedream- an unattainable or fanciful hope or plan. What else could he have meant if not to say AI one day greatly benefiting humanity is unattainable or fanciful? This seems like the logical meaning of his statement.
Tinfoil hat incoming. I feel like all of this godawful AI use is being pushed so hard just to enforce the negative view towards AI as a whole specifically to try and ruin the genuinely good, but unprofitable or difficult to capitalize on, uses of AI.
My tinfoil hat is the opposite. Keep pumping out the feel-good stories about positive uses for AI to try and change the conversation away from just how dystopian (and not even the cool dystopia) the world techbros are trying to create with the AI they're pushing on us all so hard. Generative AI has consistently been a money pit since this all started, so they have to keep pointing to legitimate good uses for AI to justify bleeding everyone dry.
To what end? Who benefits by preventing AI from being used in places where it's hard to monetize? Like cancer research or something. How would Big Pharma (I assume they would be the culprit here) benefit from making cancer drugs cheaper and easier to research?
Full cures I would understand, as then they lose money from people paying for treatment. But they still have reasons to discover those cures first, if only to bury them.
I mean, I feel like you kind of answered your own question. Cancer treatments being cheaper is less profitable for Big Pharma. That would be their purpose of sabotaging the thing being used to improve treatments.
The biggest issue with Capitalism is if the number doesn't go up infinitely it's bad. Many companies don't care about the long term, only the short term.
Also, isn't it more likely that current AI is just Capitalism as you aptly described it here? Companies pumping it out trying to make $$$ now with no concern about long term goals?
Your description of Capitalism's largest issue directly contradicts your theory. People are pumping out garbage AI stuff because it makes money fast. Big Pharma would also jump on using AI to reduce research costs and increase RoI, and to beat others to drugs so they can patent it and either hide it or profit off it.
I guess the follow-up is how does poisoning the well on AI actually negatively effect something like cancer research? Researchers are going to use tools available regardless, and using AI to analyze compounds or whatever would be like stage 6 out of the hundreds of stages to create and approve a treatment. How does it harm the research?
Um, look at the last century of technological advances. Making things cheaper to PRODUCE does not reduce the price the consumer pays. It just puts more money in shareholder's pockets.
As a research tool the current AI is a game changer. If you want to find patterns in large controlled data sets, it can save huge amounts of man hours. The issue is the uses based purely on greed.
While I can appreciate the idea, it is also the antithesis of my current job role in some ways and therefore I must distrust it on principle. A tool that looks up information, but has the potential to hallucinate, is just risking getting sued when someone follows that hallucinated information.
Siting sources is great, but only if the user will always check those sources. In theory researchers should always check every source, but they don't in reality because everyone is lazy sometimes. As long as most people are acting professional most of the time then the errors get picked up and fixed eventually, but AI does not have that process.
You can always have it output sources with its responses, but it would be up to you to verify the veracity of those sources, just like if you used Wikipedia to help you research.
Exactly! Such a tool would be a Wikipedia search, but anything marked [Citation Needed] would simply not exist. Everything the bot said would have to come from somewhere, and being able to look up that somewhere would make it a great tool. A hallucination would be easily detected because the sources would be something like a report from the World Wildlife Federation, a report from the National Park Service, and a random Geocities page that happened to contain the word "wolf".
Tyggna
I was working on this 7 years ago or so. Either no one was interested or I just genuinely suck at sales and marketing so I gave up on it. Might have a buyer now so I'm updating the code base
Redyls
AI: here is a picture of you if you were an astronaut!
ChicagRealt0r
I want both of those things.
teardropivyyearofthetiger
I was AI to price fix ALL groceries at a specific low profit level (2% tops) at the grocery store NEAREST to me, keep the jobs at the store sustainable, balanced, and the store’s property & business taxes to be set by the city by AI, so their tax level is sustainable for maintenance on sewers and roads.
I want AI to catch and punish corruption. I want corruption to be excluded from pardon (effectively treated like declaring bankruptcy) wherein the corrupt no longer able to commit corruption.
Mediocreclient
we desperately need to stop lumping all the weird shit under the "AI" label. Chat gippity is an LLM. Image generation is stable diffusion. machine learning and neural networks have been around for decades and have genuinely changed the world (sometimes for the better). they all apparently get called "AI", and a lot of "AI" fans transfer credit from ML/NN and give it to LLM/SD, stolen valor style.
technicalfool
If you're giving a computer some appearance of intelligence, it's AI. Videogame AI doesn't necessarily involve much more than some pathfinding algorithms and a set of weights for aggro, but nobody has a problem calling that AI. AI is the umbrella term. LLM, SD, ANNs, all types of AI. Hell, a nested set of IFs can be AI.
It's really not worth getting a whole rustled set of jimmies over.
Mediocreclient
not realizing that the average person doesn't grasp the technical nuance of neated terms is literally part of the problem, so thank you for highlighting the discrepancy between "a bunch of boolean strings is technically AI" and the average redditor genuinely perceiving the usage of the word "AI" to literally meaning thinking machines with souls and feelings who can actually talk to them, and aren't just churning out high-rank data nodes.
YeastInfectedWhiskerBiscuit
I just want the OS1 from the film HER. I just want someone to talk to. Sadly, we'll never get to that point. All AI being built is strictly to spy and gather information to exploit people for profit to make companies rich. Not for the human experience. Not to be helpful, or to improve our lives in any meaningful way. Just spy, profit, manipulation, and control of the masses.
YeastInfectedWhiskerBiscuit
The closest thing we have to "HER" today, is Nuro-Sama. An at-home project by a streamer that isn't made to be a personal thief for profit. Just a human-ish AI with a semi believable personality.
ameranthe
I asked chat GPT to go to a particular website and find me five meals that had to meet certain criteria and have overlapping ingredients. Then I asked it to make me a grocery list. It did really well on the first part, but it left out about half of the needed ingredients for the recipes, I think it only gave me the overlapping ingredients for some reason. I'm going to keep playing with the prompt but I'm almost there.
fractalsphere
AI/robotics is for hazardous work in the mines, repetitive motion in the factories, maybe product inspections, shipping route calculations, NOT art, music or creative writing. Those things are what the HUMAN MIND is for.
SpaceballsTheComment
Nobody's taking art away from the artists.
MaleProstateMilker88
Literally you are. All the artwork of every field was scraped and put it into the gen ai programs. None of that belongs to the techbros or the equally exploitative userbase.
SniderThanYou
I want all those things, AND a picture of me as an astronaut.
SloppyPotter
you as an astronaut
Imheretostorenottopost
I'm pretty much over "AI".
It doesn't do much useful at this point and its capabilities continue to be monumentally exaggerated.
criticalmassdriver
I just use AI to rewrite what I already wrote. Because my autistic ass writes something and it's filled with parenthesis and extra thoughts and overly detailed information and run on sentences so that neurotypical coworkers actually read the entire thing instead of giving up a paragraph in.
Copperbrat
Honestly sounds like something AI would write. Literally no one needs AI to anticipate what they need to restock, and they certainly aren't going to drive to 4 different locations for the best price on several different items. If there are people out there that think this is a better use for AI (and the resulting waste of resources), humans have become a race of idiots.
JinxRocks
I kind of use it for this very reason. I ask it to limit itself to three stores near me and ask which has the best price on size/version of product X. sometimes it doesn't always work and I have to check manually, but.. it has cut the time in half of doing the entire grocery list.
azarza
what would you pay for a way better service for something like that? or would a freemium thing be better? it's a good idea
aslum
Just don't use GenAi/ChatGPT/LLMs. Even if you don't care about the ecological harm or hallucinations, they're all built on Theft!
aslum
And don't come at me with "Oh, but X is also bad for the environment" or "X is also built on theft." Two things can be bad at once.
ReelPoop
both are doable
MoistOwlette23
AI: Here is your weekly shopping list.
List: Half the items are from Portland Oregon and half from Portland Maine.
manystripes
If you wanna get the best prices you need to be prepared to visit multiple stores
TheWinterBagel
Your whole shopping list completed with just two stops!
seekingdaedalus
In reality, it just wouldn't do what OP asked for. It probably would even seem like it works pretty well... except... it would just give you a list of sponsored products. Because tech companies don't build products anymore. They build facades that extort money from other companies.
dreammer243
Yeah, they're starting to skip the step in the enshittification process where it's supposed to be good and useful at first.
Alistairetheblu
Also it suggested glue to make a pizza and piss to drink because you looked up kidney stones that one time.
abrazenfool
Ivermectin !!!
elbowdeepinastupidsexycentaur
lucky you it picked actual real locations, get ready for eggs from portland Cyrodil
ProbablyWrong524
What kind of animal did these Cyrodillic eggs come from? I don't remember seeing a single chicken in Oblivion. Do they import from Skyrim?
FeedTheNachoMan
Clannfears
johnxbear
Somehow a third half is from a portal to another plane.
cdlong
Do they deliver? It seems like shipping would be convenient and fast. (Or an unbelievable pain in the ass)
zenoshogun
Now you're thinking with portals!
keillrandor
The portal to Spain falls mainly on the plane.
onlyhalfghost
ah, the SCP fourth Portland, accessed from Portland, Oregon; Portland, Maine; and Portland, in the UK. >
onlyhalfghost
> https://scp-wiki.wikidot.com/three-portlands-hub
eggmuffin
Portland, Portal-land, whatever.
yzark01
Thats funny youre assuming it would get the list right?
StrangePaegan
And at least three items will kill you.
voltus
No no no silly goose AI is for making corporations more money not saving you money.
moonshadowkati
Doable. Go ahead and build it. :)
pmather021
That's called auto renewal. You want a grocery store inventory system.
Atomic2
But how do you expect it to anticipate that without studying your habits? People are reluctant to feed more data to the machine but this would require a LOT of data. On some level, you'd need to be okay with a machine observing you.
vorodar
Camera in the fridge, use ML to identify what products are there. You can set up stock limits (e.g., "12 eggs", "2 bottles of milk", etc) and when requested, it produces a list that fills up to meet the limit (2 eggs left = 10 more needed, 1 bottle of milk left = 1 more needed, etc). It doesn't have to be complicated or manifest 1984.
Atomic2
Yea but that is exactly my point--you'll need to be okay with a machine observing you and creating a profile on you. also a single camera would never capture everything in a full fridge. It would have no idea how many eggs are left in a carton. How much sour cream is left in the tub. Etc. It can infer, by how many times its been removed and compare to historical data, but never truly know.
TheElliotPage
LOL actually, my Samsung fridge has done that for years
HumanCats
"I ordered six fishfingers but got sent nine".
"Yeah AI does my shopping".
rIOsK
Generative AI is just a branch of what we colloquially call AI
clonedeeznuts
You're actually thinking of real world problems and technology to solve them vs the broligarchs creating products without a problem to solve
Grenateh
Literally though this is what AI should be used for.
L34TSP3k8888
Very few business models exist to save citizens money and help corpos make less. Wonder why.
BlackBearChopSticks
Instead they’re using AI for ‘dynamic price tags’ that are connected to Wi-Fi, have a camera for your face, and will profile you on the spot for the price. It will also raise prices to scalp to the it’s best ability.
SomeDetroitGuy
LLMs will never solve that problem. It isnt at all what they do.
McFrazzlestache
Sounds like you just wrote an app. Develop and market it, so you can collect piles of gold.
Sylventhe
But what happens when someone sends an army of Orcish Grandmas to raid your gold piles? Wait, I think we have two different apps in mind here...
johnxbear
Tyggna
I wrote that app 7 years ago, couldn't get as single person to try it as I suck at marketing. Gave up and went back to the 9 to 5
B99Reactions
Just post it on imgur, say you quit your job to do it full time, make some sob story and boom, you just got another ad through imgur front page
Tyggna
/gallery/i-made-thing-4vQOJ Here's the post that died in usersub from it. It had vision AI and could already correctly id items in your fridge. The theory is that farm-direct would always be the cheapest option if available. I got a restaurant food distributor interested in it two weeks ago, so I'm updating the code base and working on ERP integration. I'll just keep tinkering with it so long as I stay unemployed.
B99Reactions
Weird it died in Irvine, I guess sometimes they make it but sometimes they don't
B99Reactions
Usersub*
Californiajackson
Hmm. Well, here we are. Your test subjects await. And I'll bet someone on here excels at marketing.
3Davideo
Just because I can state the problem doesn't mean I currently have the skills to build an app capable of solving the problem.
McFrazzlestache
No, but, collaborators do, and they turn into partners. Think of it like commissioned art. You know what you want to see, just can't put it on paper. Artist can put it on paper, just needs a blueprint.
Malloon
Also money. You need either money, connections or skill to get such a partnership. A mere idea won't get you anywhere unless you patent it, in which case it falls under skill.
McFrazzlestache
Excuses don't get sprinkles. Only winners do.
Feralkyn
Commissioned art, you say? Why does this coupon-searching AI suddenly have giant titties, and why is it a fox?
McFrazzlestache
Graded on a curve, and most AI bots are stacked furries now for some reason.
Druidhunter77
Everything becomes stacked furries eventually. It’s like evolution and crabs.
jeejeejerrycotton
That's not actually a job for "AI", that's just the perfect job for regular old-fashioned ADP (Automatic Data Processing. What we had before fuckos and goobers came up with "IT".) The reason we don't have it is because capital owns the world.
VodkaReindeer
Oh cool, I didn't know there was an English world for ATK.
Arcanum3000
Nah, the reason we don't have it is because there's not a good way to automatically track how many eggs you have left and the fact that there's only a half a glass of milk in the jug.
Hendlton
And also because there isn't a good way to track what's available where and how much it costs. Most stores around me don't have all their items listed on their websites. And even if they do, they don't specifically list what's available at the one closest to me. I don't know how AI would fix that, but AI can certainly look at a picture of a fridge and list what's in it.
vegivamp
Funny you say that. I can't put in a collect order without selecting the shop I'll be picking up in, and not just for the obvious reason but also because prices differ regionally, and because availability differs between shops.
Grenateh
IT has been a thing since before you were alive.
jeejeejerrycotton
Eating poop has been a thing for longer than I've been alive. It hasn't been very popular, however
myotherusernameismyotherusername
Germany has entered the chat.
Isorikk
How old are you? ADP has been a part of IT since the 1940's
potshot
Traditional automated data processes all rely on strictly defined data formats and sources. The thing with AI isn't that it does the same thing traditional methods do, it specifically has an edge on not requiring a specific format. It's absolutely not something that was doable with "ADP".
IconicM
Well, it could be done but wouldn't really be worth the effort of manually figuring out how to scrape every grocery store's website...
potshot
Yeah, kind of, but in this case I meant "not doable" as in not actually feasible as a product that people can use. It's specific to the person, or at least the neighborhood, it takes a lot of manual work to configure it for each shop you want to include, it likely needs a lot of maintenance, and each use-case has to be separately defined. With AI, you can skip all that. Today you want to know where to get baking products, tomorrow you're interested in gardening tools, and next week you need
potshot
to replace your jeans? an AI system designed for that can handle it, no problem. But with traditional methods? Each shop has to be painstakingly added and processing methods implemented. Not to mention any kind of intelligent system that tracks what you need(this doesn't exist yet, but it is feasible with AI agents).
potshot
Now to be clear, I don't think you should use AI for this because of various legal and ethical reasons, and I also don't believe that the current LLM/LRM approaches are not ideal for this kind of daily task assistance work. But the fact is, they do make it feasible to implement a system that works with enough accuracy to actually be usable, as opposed to traditional methods that simply aren't doable within a moderate amount of effort from the user and are absolutely not scalable.
zafner
Make the app
Make the app
Make the app
Charge $.50 or something. Don't solicit advertising. Everyone would do this
CyberHexx
Yes, the process is Automatic Data Processing, but the input output device I want to initiate that ADP is a simple assistant, to whom I can say, "Check my pantry, and make a shopping list. Prioritize least number of shops with max number of sales, coupons, and deals."
lurkertron5000
*capitalists. With names and addresses.
lurkertron5000
Shit, someone who thinks an abstract term literally owns the world downvoted me. I guess no one owns the majority of the world's resources?
VaultGirl69
So many things are marketed as AI. I've spent 20 years in tech and it's just silly to watch. Especially those naive regarding what it can and can't do, or believing it understands just because it can regurgitate things it's taught.
cepacolusmaximus
turns out, AI is really good at data processing, and its natural language command structure makes it accessible to many, instead of just programmers
Pilgrymm
Hey don't disrespect IT workers, I've been IT for years. Ai suck but don't blame us for it.
alcamar
IT says fuck AI, at least those of us who know what stupid shit it really is
Davidnfilms
Amen brother
yzark01
Yep. Not a single other engineer ive talked to has anything good to say about it. We all realize its an overhyped predictive text run by statistics that cant really do anything well except help with writing. Every non technical person ive spoke to thinks its magic sentient machine that actually knows things.
We're fucked because no one can be bothered to even watch a 30m video on youtube explaining how it works.
Pilgrymm
Truth. I hate ai. It's why I keep a gun next to my printer.
Illumian85
.... wouldn't it be better to keep the firearm out of the printers reach?
Flobmmando
It's not a place to put it, it's a reminder for it to print as I tell it too.
Pilgrymm
I chopped her hands off, duhh.
I mean c'mon, pssh, I've been doing this for over 20 years now.
alcamar
Unloaded so you know when they try and fail, so you can load it and put the printer down before it spreads further
DonjuanV
Im pretty i could do this with power automate. Maybe not the search for best price part
ravnicrasol
The reason we don't have it is because the AI the post describes is a data-privacy nightmare.
"Yes, please, I would like for you to keep track of everything I purchase and own, my eating habits, geolocation, and purchase habits. I'd also like for you to decide and shove 'coupons' based off of that information. Also also, please know literally every store, deal, price, and location in the region, and decide for me which will better fulfill my needs."
Heninthefoxden
Honestly like this already happens Tho. I was able to link my kroger savings account to instacart to get the same savings benefits. I enter my phone number wherever I shop these days. I like it because with modern digital tracking, you can pretty much find and verify every single transaction. If there's a financial problem or a product problem it's very easy to track.
nemesisx00
And why is privacy so vital? Because capitalists will exploit every possible perceived weakness in order to extract more money, an imaginary resource, from everyone else on the planet with no regard to the real harm they do in the process.
FeedTheNachoMan
I mean, personally, I don't care so much about the deals part. If I had an AI assistant, I'd want it completely air-gapped from the internet.
But also I want it to remind me to actually bring the reusable grocery bags when I go shopping.
FeedTheNachoMan
And no, I don't use the fuckin Copilot or Gemini bullshit on my computer or phone, so I don't count that as having an AI assistant
BJWTech
Best to run your own LLM then.
BishlamekGurpgork
You typically have very poor results unless you're spending a lot of money.
vegivamp
Matter of time. Your old Nokia 3310 had orders of magnitude more computing power than what we went to the moon with. Your current PC blows turn off the century supercomputers out of the water.
In 10 years, a current computer will be able to run a pretrained LLM of today's quality, no problem.
BishlamekGurpgork
And that helps Ravnicrasol today, how?
IrrelevantIrrelevant
It's comical how many people will, in one breath, hate that AI is being used to gather personal information, and then in the next wish that they had an AI that knows not only their own information, but also all the information for the surrounding few towns so they can get deals.
Not to mention, the type of AI that does that has nothing to do with generative art. It's completely different people making completely different products, but people just see the word "AI" and think its all one thing.
vegivamp
I don't have a problem with AI gathering information to be useful to me. I have a problem with that AI gathered data being used and sold.
ElbowDeepInAPoliceState
The underlying software for route planning and optimization and all that probably wouldn't be done inside tge LLM, however searching fliers and the like is absolutely within the realm of possibility for one, as is using one as an 'agent' to perform all the other clerical work that would go into something like this.
LLMs are extremely flexible due to the fact that 'language' can be any means of information encoding, afyer all they're predictive algorithms, there's no real comprehension.
Hammerwell
If it were an AI it would be the same. The same a John Doe searches in his citys shops informations for the best deal and paints a picture in the evening. Otherwise its not an artifical intelligence but an artifical stupid.
ravnicrasol
Intelligence: the ability to acquire and apply knowledge and skills.
AI is, in fact, definitionally correct. It's an intelligent tool (one trained to solve specific sets of tasks. Be it predict text, how proteins fold, or draw tits)
But people seem to have this weird habit of thinking "if intelligent, then needs to be intelligent in all things". As if nowadays we don't have plenty of examples of people who're geniuses in one thing and morons in another.
Hammerwell
Self awarenesses would be the definition. And that would make the current use slavery.
IrrelevantIrrelevant
On top of this, AI is still in its infancy, in all current variants.
People are basically asking "Why can't my baby walk yet, what good is a kid who can't walk?" when it just barely made it to toddler status. We just barely taught the thing not to shit itself, needs time before it becomes capable of true greatness.
The fact that grifters are busy making money off of its current variant is a testament to the giant it is going to become as people work out the kinks behind the scenes.
AyatollahBahloni
https://garymarcus.substack.com/p/a-knockout-blow-for-llms Good article.
"Whenever people ask me why I (contrary to widespread myth) actually like AI, and think that AI (though not GenAI) may ultimately be of great benefit to humanity, I invariably point to the advances in science and technology we might make if we could combine the causal reasoning abilities of our best scientists with the sheer compute power of modern digital computers."
TheElliotPage
meanwhile...
IGuessImAFan
Yeah check out the advancements in protein models, AI is definitely doing good
cambam3000
I would also consider how many jobs are being lost to AI right now. I am constantly triggered by seeing Republicans talk about "illegals stealing our jobs" while literally AI is openly doing it for everyone to see. I'm probably going to be retraining for a trade because of that, and it's not making me a happy camper ;.;
adiving
I technically agree that would be awesome. But considering current AI is nowhere even remotely up to that point it feels a bit weird to defend current AI with this pipedream.
NothingAndAHalf
The article is not defending current AI — it's about why LLMs are _not_ going to be able to do this. The above quote is not a good summary of the article. Another quote, from the article: "But anybody who thinks LLMs are a direct route to the sort AGI that could fundamentally transform society for the good is kidding themselves."
TheElliotPage
last year, they were training AI & after only 3 or 4 days, the AI started training other AI that "humans must be destroyed" ...yikes...
sanguium
It is, look up AI on protein folding, chatbots are just the lowest form of current AI, wasteful unsustainable and innacurate, the fact that they are the face of ai in the general public is just a marketing stunt for openai to keep burning billions, yay capitalism
senseicombs
So technically you agree it would be awesome but you believe it's a pipe dream? So are you saying that you believe it will never reach a point where it could be a benefit to humanity? And if so you must then believe no human will ever be able to make AI functional to the point of achieving this goal. If that is what you're saying you must believe you know more than all the people working on advancing AI that do believe they can accomplish this goal. Do you know more?
adiving
Holy fucking assumptions batman.
senseicombs
So are you saying the pipedream you're referring to is not that AI could, "ultimately be of great benefit to humanity"? It sounded like you said that was a pipe dream to me. What was the pipedream you're referring to if I made an assumption? It seemed like that's what you said to me and was therefore not an assumption. Maybe you can explain what you meant using words from the context in which your reply was made?
fnxweb
I saw it as: seems odd to like AI for something it isn't [yet, if ever], which is what the quote is implying
senseicombs
He said "It's weird to defend current AI with this pipedream" The pipedream, as defined in the statement, that AI could one day be of great benefit to humanity. Pipedream- an unattainable or fanciful hope or plan. What else could he have meant if not to say AI one day greatly benefiting humanity is unattainable or fanciful? This seems like the logical meaning of his statement.
BlueDsc
Tinfoil hat incoming. I feel like all of this godawful AI use is being pushed so hard just to enforce the negative view towards AI as a whole specifically to try and ruin the genuinely good, but unprofitable or difficult to capitalize on, uses of AI.
Bobothehobomcjojo
My tinfoil hat is the opposite. Keep pumping out the feel-good stories about positive uses for AI to try and change the conversation away from just how dystopian (and not even the cool dystopia) the world techbros are trying to create with the AI they're pushing on us all so hard. Generative AI has consistently been a money pit since this all started, so they have to keep pointing to legitimate good uses for AI to justify bleeding everyone dry.
LiminallyInsane
I wish. No, it's just short-sighted capitalism ruining things that could actually benefit society. Again.
Badprenup
To what end? Who benefits by preventing AI from being used in places where it's hard to monetize? Like cancer research or something. How would Big Pharma (I assume they would be the culprit here) benefit from making cancer drugs cheaper and easier to research?
Full cures I would understand, as then they lose money from people paying for treatment. But they still have reasons to discover those cures first, if only to bury them.
BlueDsc
I mean, I feel like you kind of answered your own question. Cancer treatments being cheaper is less profitable for Big Pharma. That would be their purpose of sabotaging the thing being used to improve treatments.
The biggest issue with Capitalism is if the number doesn't go up infinitely it's bad. Many companies don't care about the long term, only the short term.
Badprenup
Also, isn't it more likely that current AI is just Capitalism as you aptly described it here? Companies pumping it out trying to make $$$ now with no concern about long term goals?
Your description of Capitalism's largest issue directly contradicts your theory. People are pumping out garbage AI stuff because it makes money fast. Big Pharma would also jump on using AI to reduce research costs and increase RoI, and to beat others to drugs so they can patent it and either hide it or profit off it.
Badprenup
I guess the follow-up is how does poisoning the well on AI actually negatively effect something like cancer research? Researchers are going to use tools available regardless, and using AI to analyze compounds or whatever would be like stage 6 out of the hundreds of stages to create and approve a treatment. How does it harm the research?
Zaranthan
Um, look at the last century of technological advances. Making things cheaper to PRODUCE does not reduce the price the consumer pays. It just puts more money in shareholder's pockets.
PirateRubberDuck
As a research tool the current AI is a game changer. If you want to find patterns in large controlled data sets, it can save huge amounts of man hours.
The issue is the uses based purely on greed.
Zaranthan
If someone would make an LLM that cited its sources, I'd defend it to the ends of the Earth. That would be a magnificent research tool.
PirateRubberDuck
While I can appreciate the idea, it is also the antithesis of my current job role in some ways and therefore I must distrust it on principle. A tool that looks up information, but has the potential to hallucinate, is just risking getting sued when someone follows that hallucinated information.
PirateRubberDuck
Siting sources is great, but only if the user will always check those sources. In theory researchers should always check every source, but they don't in reality because everyone is lazy sometimes. As long as most people are acting professional most of the time then the errors get picked up and fixed eventually, but AI does not have that process.
RatsLiveOnNoEvilStar
You can always have it output sources with its responses, but it would be up to you to verify the veracity of those sources, just like if you used Wikipedia to help you research.
Zaranthan
Exactly! Such a tool would be a Wikipedia search, but anything marked [Citation Needed] would simply not exist. Everything the bot said would have to come from somewhere, and being able to look up that somewhere would make it a great tool. A hallucination would be easily detected because the sources would be something like a report from the World Wildlife Federation, a report from the National Park Service, and a random Geocities page that happened to contain the word "wolf".