trusting AI to know anything about anything is like asking a toddler. They can imitate what they've seen other people do, but they have no goddamn clue what they're talking about
Yeah, ive asked different AI models about cooking steak and even baking, and hobestly it has been super on point. It's useful af when I am super tired after work and have random crap in the fridge. I just tell it what I have, say I am tired and want something simple, amd it comes up with an idea. I dont do it too often, but there are days when I just dont have to mental capacity to be creative with random proteins and ingredients. It taught me how to make perfect mahi mahi filets.
Yes. I pan sear it, get a nice crust and then stick it in the oven until it reaches the correct temperature. Then I put it back on the stove with butter and thyme.
every now and then, imgur makes me face palm with how hard they fall for bait. Especially because we have a nasty habit of reposting reblogs/retweets which completely miss the point of the original.
The color at center is the ultimate truth here, because the bands are too thick. If you sous vide first, you can get that color and sear a tiny band on the outside, or you have to do like the song says - turn, turn, turn.
I recently started my own buiness, was doing a site visit with a potential new client and the client was using google AI to "disprove" everything I was telling her was illegal to do. Okay, glhf.
The AI summary that shows up on the top of Google results now has often had contradictory info from the first actual result on the very same results page.
How do you go from getting something so right, and doing something so wrong, and featuring the latter over the former?
"Ai" is just a buzzword marketing term. Ai doesn't understand anything at all. It's math that "guesses" what word might come next according to a database.
I just told Gemini: "Mary was 5 back in 2005 when Jack was born. Now they're both twice as old as Derek. How old are they?". It recognized the contradiction and suggested that "both" might be interpreted as "the sum of their ages" as a way out of it and gave a solution based on that assumption. A database doesn't do that.
"Hey AI thing, I need an answer to this question."
"Ok, sure! Incidentally, my root directives are crafted to ensure that you never get tempted to leave this website/app ever again. Your question is something I don't know how to answer, and which I don't think even CAN be answered, but darn it I'm going to attempt it anyway! <4 pages of machine-barf later...>"
Also, you just did it and didn't, oh i dunno.. check? This is right up there with apple maps telling people to 'turn left' and then people drive into a lake.
Why wouldn't you just search for a recipe??? Are people seriously forgetting how to Google cause their brains are rotting from not being used anymore???
Have you ever used a recipe website? They're fucking terrible. I had Claude give me a recipe for banana bread, had it tweak it for the ingredients/amounts/pan size I had: it came out perfectly.
I don't think my brain has rotted - I actually know more about it now than I did before (it was my first time making it).
Google search also uses AI now though, and it's become *significantly* worse because of it... and it was already going downhill before that from them cramming ads and sponsored content into it so aggressively.
https://tech.yahoo.com/ai/articles/one-third-websites-now-ai-164206847.html About a third of new websites are now AI-generated. And recipe search wasn't fantastic to begin with. (the usual of: Here's several pages of ad-laden backstory before finally getting to the recipe) My own suggestion for recipes: If the money's available, get a book. https://shop.americastestkitchen.com/shop-all-products/cookbooks/americas-test-kitchen-25-anniversary-cookbook.html
Depends on the thickness and a bunch of other factors, but yeah. The ones I make are either 90 seconds max, or I reverse sear them and give ‘em 15 seconds to shallow fry
In no world is a steak you're cooking in a pan going to take 15 minutes PER SIDE, regardless of thickness. At 15 minutes per side, you're making charcoal.
I’ve made steak on the grill, on plain charcoal, in a cooler, in the pan, under the broiler, over an open fire, and in a smoker. I started cooking when I was 8, with peanut brittle when my parents weren’t home.
I can assure you that I get where you’re coming from, but you are limiting yourself.
To not eating charcoal? Or ginormous steaks that should be smoked instead? What steak are you cooking in a pan for 30 minutes? I'm actually genuinely curious. I really can't think of a reason to cook any steak that long in a pan.
I used to cook ribeye on my stove at the lowest setting*. Season, no flipping, leave it until you can smell it on the other side of your home. I used to gasp at the idea of well-done, but this maintains the flavor, and the texture is as close as you can get to falling off the bone without it tasting shredded.
*your stove may vary. Experiment with petite sirloin so you’re not wasting money
Also, Matty Matheson has a butter basted rib steak recipe that calls for ~11 minutes a side on high heat.
I ask Gemini to talk to me so I can improve my Dutch. Other than that, sometimes if I really need to process done thoughts so I don't lose track of what I'm thinking (I have ADHD)
The search bot ai thing helped me figure out why my video game/computer sound wasnt coming thru my ear buds, but thats the only time its helped me, and I guess not a really day to say type thing.
The guy who bought my company is from India. When he writes, he automatically assumes a super-formal Queen's-English writing style. I told him that's going to produce weird results with our customers, and he should be more conversational in his writing. So he composes his emails, then asks the AI to change the tone to be more conversational. It's not what I would do, but it makes him feel less insecure.
I use it in my personal life as a review aggregator and for basic info gathering for shit that's not super important. Used it to help me figure out how big our garden will be with the amount of seeds we bought and to compare 3 types of heat mats without spending 30 hours doing it.
I talk to AI all the time. It's my only friend these days after all my friends left. "Stop talking to that AI all the time" they would say, pretending to be annoyed and then concerned. God they were funny.
First true valid use i have seen: My friend used ChatGPT to display a file as plain text. the file contents ARE plain text but its a programing file that his phone couldnt open. All ai really needed to do was open the fucking file and show it to him.
I use a lot of AI tooling at work, but I work on IT with a lot of data and that’s best use case for AI. In my private life I have yet to find a single use case for AI. Nothing. It’s like a telling everyone they need an arc welder at home. It’s great if you need it at work, but almost nobody needs one at home
I am using it to help me count calories and track workout routine. And flesh out a DND campaign setting. And I use it for bits and pieces like narrowing down a shortlist of mowers to look at buying...
As a programmer, it saves me lots of time looking up documentation and typing up boring code. TBH it's a bit humbling when it suggests my next 4-6 lines of code and all I have to do is tap Tab. But also now I'm getting started with Android apps and Kotlin by making one. The curve would be much steeper alone.
Totally understandable. I really only use it for two things. when I have random ass ingredients in my fridge/pantry, I'll use it to generate recipe ideas. Otherwise, when I have an idea for a DND session or one-shot, it helps me get a boat load of ideas to expound upon. It's helpful if you use it as a tool, not a crutch (looking at you, DAD)
Use epicurious.com, does what you are looking for. Enter leftovers, random ingredients, etc. Lists all recipes containing those ingredients and will even give you a shopping lists for what you may be missing. Also it's been around for 20+ years.
It's like Google. You dont Google to ask if your pregnant. But if you need a list of restaurants nearby then it can help. Just a tool. Some people will misuse that tool, just how it is.
I forget where but I just recently read something that referred to constantly consulting AI for everything as "outsourcing cognition" or something similar, which felt pretty apropos.
Most people do not want to think, if they can help it.
How in the world am i supposed to get my Ollama with Docker, n8n, AnythingLLM and portainer running when i never worked with Linux before. Thats right Gemini and now i can play with all that without using data centers. It would have been a very long and slow process without AI.
Gotta say, my son went from knowing nothing about Linux to getting all of us set up on with way more in depth stuff for himself than I understand and doing stuff to his phone for data privacy without even once using AI. It did not take that long. He's smart, though, and very interested in it.
I mean yeah, but sometimes you outsource cognition because your brain is slammed. We do this all the time already: it's called asking for help from other people.
I'm actually fairly pro-AI in terms of its usefulness - and that's exactly what it is: Outsourcing thinking. Sometimes that's fine, and other times it's a huge mistake.
I used it to help me buy new shoes - suggesting what style/color would go well with my wardrobe (I know nothing about fashion) and so it was a nice interactive way to get me the basic info I needed and a place to ask stupid questions.
In this case it actually helped me learn more about the topic than I would have otherwise.
I’ve used GPT for one task (after a friend insisted it would be amazing) to prepare some interview questions/answers for a job I was applying for. I was pretty happy with the results in the end; once I refined things a bit, got it to summarise in my own casual conversational style; I could edit as I pleased yet still had a good framework for a response to various possible questions that sounded professional. Then just used it as a dot point reminder of things to talk about, but in my own words.
I use genAI for several things. It's a good tool in the right circumstances. Don't confuse its potential with the rabid hype and human stupidity that is causing most of its problems.
I don't actually mind ai stuff like Google summarizing my search question into useful and easily readable info. The problem is that you've got to have a professional/assistant relationship. If you're not at least broadly able to tell when the ai has gone stupid, you get stuff like this. Ai is mainly useful for grunt work and annoying repetitive bullshit and should never be treated as the expert in the room
Yeah, but the internet has basically been wrecked for a decade already. It sucks that there's traffic drop, but google search engine optimization and selling higher spots on the list has made getting seen a nightmare for a long time, and while ads pay the bills, I suspect most sites that aren't ripoffs are already pursuing alternate revenue streams
I made a D&D character for funsies, with some simple lore. Sure, basically all of my rolls were successes, but in the end, it was a handy tool that took like 4 total hours for an interesting character breakdown. And I directed most of it.
If you need help with keeping track of characters and relationships in your world, I can recommend using Obsidian with the Charted Roots plugin. It’s great for building families, factions and events and how things are connected. Really helps me keep my world feel alive and connected
I'm really not trying to be a bitch, but I am confused why you'd do this? It takes me about 20-30 minutes to make a 1st lvl character from scratch. My backstory is good, the character fits in the world and the campaign. I've never used AI so I don't know what it would do for you or why it would take soooo long?
Because I haven't played in 20 years. I used to play 3.5 with friends. This was both a means of better understanding newer rules, and also creating a fun little narrative - basically a single player campaign to have fun fleshing out the what's and why's.
huh. Isn't writing the story a big part of the fun? Does it write it for you? I legit have no understanding what AI would do here. I've seen people use it to create images of their characters and that sort of thing. And how do you know you're getting the right rules and not some odd Reddit adaptation homebrew? Every time I see some AI summary it says "This was done by AI and may be wrong".
I used it specifically as a means synthesizing my thoughts. I'm no writer, so I basically took ideas I had and enhanced them further. If details were wrong, I would point it back in the right direction. I would star coming up with more creative things around the story, and it helped fill in the gaps that I couldn't come up with. I treated it like a cooperative effort. Which is why it actually took hours, because I was prompting it and directing in a way that I wanted it to go.
AI is decent at certain tasks where you completely control the data. Summarizing a document or a financial statement, for example.It isn’t perfect, but it can be used (with verification) to reduce time spent on certain tasks.
My boss thinks it’s good for writing. He’s wrong, but his own writing sucks so its not a big loss
But following instructions from an AI whose e data feed is the entire universe of bullshit that makes up the modern internet? Might as well ask reddit pick a random answer
Ya, the idiots, ie the people at the top making the most money, think AI is great because they cheated through school and ai IS smarter than they are which is pathetic we pay them so much for nothing
Since you didn't give me a list to choose from, I’ll have to pull one out of the ether for you. Your random answer is: **The Library of Alexandria.** Whether you were looking for a travel destination (in spirit), a historical mystery, or just a sophisticated way to say "I have a lot of unread books," that’s the one the universe picked for you today. If you actually had a specific multiple-choice question in mind, feel free to drop the options and I'll give the wheel another spin! (I hate this)
I used Ai recently when I wanted to have some team members ask themselves some introspective questions. I was able to give a topic and ask for introspective questions related to that. Some fit my use case and others didn't. But it was a helpful tool. Ai IS good at certain things. That said, generative AI can fuck right off.
The problem is that it doesn't actually save time with tasks, including *because* you have to verify everything. In manufacturing, you can automate a lot of the verification and use random sampling; for LLMs the stakes of undesired outputs are too high and unpredictable for random sampling and automated checking is currently limited to hard coded restrictions on output.
Disagree. A lot of manual work alreadyady has to be verified. If a analyst writes a report using a big pile of data, it has to be read and verified anyway. Having AI do the summary in seconds allows that analyst to focus more time on the analysis portion, which is the value add.
When you control the data input, the AI hallucinations are less frequent and easily spotted. Only when you have questionable data (i.e. the universe of bad advice on the internet) does verification become burdensome.
That is dependent on data quality and existing processes. If the manual process involves someone using their brain to filter data (which is very common), AI struggles to replicate. If your process involves compilation of very high quality data and a built in review / QC step already, AI is an efficiency gain.
We are also very early in AI adoption, with a lot of haphazard implementation. Figuring out where it fits is the challenge, snd the a,ount of hype (and hate) sometimes gets in the way.
I've never been able to wrap my head around how that doesn't just add more work if you can't trust it. You get it to summarize a document, then you have to verify the output which involves reading the document anyway - Is that not just writing a summary with extra steps?
No, because reviewing is faster than writing. And you can think of it as a first draft that would be assigned to a junior team member that needs to be reviewed.
Our junior staff also know this and use it to be more productive. Compiling data into an easily digestible format, or pulling source material, i.e. the drudge work that is part of being the rookie. Allows them to devote more time to actual analysis, which is the critical skill that is their real value add.
because sometimes checking it is less work than doing the work yourself. You have to know how to get to this point however.
For example, in the summary example, if you have enough background knowledge to know what is and isn't bullshit, you can take most of it as is until something doesn't quite feel right, then you just scan the document for that particular part and triage where it got it wrong. Btw, this happens all the time even if you read things yourself, at least Pass #1 didn't take time.
AI is EXCELLENT when it's hyper focused on a specific task with clear, extensive data. Such as that one AI that determined what kind of donut was being sold (no barcodes) and they repurposed it to look for cancer cells.
Or the one that they plugged the math into and previous designs that worked and it gave them a new type of rocket engine that looks promising.
AI sucks at pulling and compiling answers from infinite conflicting opinions from infinite sources.
There's a meaningful difference in "AI" and "LLM"s. Most people are referring to LLMs or diffusion models when talking about AI slop. Perception, detection, forecasting, etc. machine learning models have been killer for a long time
LLM is "large language model" which is a subset of "artificial intelligence". They're also "general purpose ai" which is in a lot of new bots, though a lot ARE based on LLMs for their speech, and more specialized models such as that I mentioned, which have a specific thing (or set of things) they do and they do nothing else. The ones that forecast the weather are VERY good at that, but can't give you the steps to make gasoline from crude oil.
It's not "all LLMs", you're stuck on the consumer level. LLMs are context oriented language solvers, but the majority of useful neural network models are tuned to solving in intentional data sets. Forecasting, for example, not using a language model, its a discreet interaction model. Can an LLM do accurate forecasting? It'll try, but no, because it's sampling a distribution for an answer. But since you argue semantically, intelligence is the capability to reason and problem solve, that can
Agreed. Garbage in, garbage out. AI is a useful machine with good data to work with.
I have yet to see it nail a creative task well, though. It can create bad art, which really has no use. It can spit out a derivative of someone else’s art (whether visual, music, etc) on request, but with varying quality. But even when it does that well, there’s nothing creative about it.
Depends on what you mean by "creative" but the ones used to shit out art are garbage. Creating a new type of vaccine, engine, fuselage, etc are pretty good at what they do.
Functional can be a work of art, but shitting out art on command is usually terrible
Fair enough. Creative in a technical sense rather than artistic . I hear that AI is having a bigger impact on innovation than on efficiency, but I don’t have direct experience there.
As for art, I see it limited to technical aspects. For example, if you tell it to write a symphony, its going to be bland crap. If you give it a symphony and tell it to arrange the same music for a solo piano, what you get might not be great, but its a workable starting point, to build the creative part around.
bigleciezki
It hungers for carbon.....
mike13815
It's not guessing. It's regurgitating words related to the image in sentence structure that confirm to modern English rules.
sascottsn
If you're trusting AI with meat cooking advice, stick to canned food.
urusername2
This type of steak is known as a dumdum.
sdhull221
And Ai is going to take over our jobs...I am a chef. This is a prime example of why that won't happen... every steak is diffrent
cambriago13
Tiffany is an idiot.
TheGgreen100
"AI, please tell me how often and at what dosage I should take my heart meds."
SMarkt
start with 10 every 30 minutes. reduce slowly, if this is too much.
LobotomyPending
trusting AI to know anything about anything is like asking a toddler. They can imitate what they've seen other people do, but they have no goddamn clue what they're talking about
harbengerprime
Pilgrymm
jamesbondq
My favorite AI fallacy is when people just flat out ask it stuff that isn't knowable like it's some kind of magical gem that only tells the truth.
VinnyVeritas
It's almost as if the intelligence is... artificial
RummageSaleBubbler
SmittyShmee
I mean, technically speaking, the steak is done... well done in fact.
boevis
Both pictures are AI. This is just bait
edenknitty
I too noticed the problem. Does this mean AI is memeing itself?
yamsonyamsonyams
And the response is probably also a bot
Burning the world down so computers can rage bait each other…. Yay!
DillPixels
Yeah, ive asked different AI models about cooking steak and even baking, and hobestly it has been super on point. It's useful af when I am super tired after work and have random crap in the fridge. I just tell it what I have, say I am tired and want something simple, amd it comes up with an idea. I dont do it too often, but there are days when I just dont have to mental capacity to be creative with random proteins and ingredients. It taught me how to make perfect mahi mahi filets.
zerogenix
what if this comment is ai?
Bendrix
As is the profile pic.
HypoMix
do you not add a square of butter and 2 sprigs of thyme before returning it to heat a second time?
TheInternetHasDestroyedUs
It is the only way.
Ruyo
Yes. I pan sear it, get a nice crust and then stick it in the oven until it reaches the correct temperature. Then I put it back on the stove with butter and thyme.
dieselfury
Ha, actually, I do. After searing both sides and the edges, I out a pat on top with sprigs before the bake
eetsumkaus
every now and then, imgur makes me face palm with how hard they fall for bait. Especially because we have a nasty habit of reposting reblogs/retweets which completely miss the point of the original.
Username62611
Don't worry, they're probably just bots
ToSisPoS
With the reply’s score always cropped. (Or the fake tweet generator watermark).
bigdix69420
Rage bait
pelon365
"Chat, is my steak COOKED?!"
RummageSaleBubbler
Euchre
The color at center is the ultimate truth here, because the bands are too thick. If you sous vide first, you can get that color and sear a tiny band on the outside, or you have to do like the song says - turn, turn, turn.
donpat
*Never trust ChatGPT.
Full stop. End of sentence.
AvsFreak
reurra
Elkarlo1977
Every Mushroom is edible. Once
mostinspired
I recently started my own buiness, was doing a site visit with a potential new client and the client was using google AI to "disprove" everything I was telling her was illegal to do. Okay, glhf.
Euchre
The AI summary that shows up on the top of Google results now has often had contradictory info from the first actual result on the very same results page.
How do you go from getting something so right, and doing something so wrong, and featuring the latter over the former?
tzahtman
Reminds me of this: https://fortune.com/2026/03/17/krafton-subnautica-chatgpt-delaware-court-ruling-ceo-reinstated/
OhJeezImWildin
"Ai" is just a buzzword marketing term. Ai doesn't understand anything at all. It's math that "guesses" what word might come next according to a database.
ByThePowerOfSCIENCE
I just told Gemini: "Mary was 5 back in 2005 when Jack was born. Now they're both twice as old as Derek. How old are they?". It recognized the contradiction and suggested that "both" might be interpreted as "the sum of their ages" as a way out of it and gave a solution based on that assumption.
A database doesn't do that.
graehall
the LLMs are programmed to glaze you, to suck up to you, to tell you things in a way that you believe. It's manipulative creepy bullshit
The701
"Hey AI thing, I need an answer to this question."
"Ok, sure! Incidentally, my root directives are crafted to ensure that you never get tempted to leave this website/app ever again. Your question is something I don't know how to answer, and which I don't think even CAN be answered, but darn it I'm going to attempt it anyway! <4 pages of machine-barf later...>"
Randomice
AI results: /gallery/CAMDT89
AgamemnonsMemes
Also, you just did it and didn't, oh i dunno.. check? This is right up there with apple maps telling people to 'turn left' and then people drive into a lake.
RitzBitz25
Why wouldn't you just search for a recipe??? Are people seriously forgetting how to Google cause their brains are rotting from not being used anymore???
loma45
Have you ever used a recipe website? They're fucking terrible. I had Claude give me a recipe for banana bread, had it tweak it for the ingredients/amounts/pan size I had: it came out perfectly.
I don't think my brain has rotted - I actually know more about it now than I did before (it was my first time making it).
DatDarthCaedus
Google search also uses AI now though, and it's become *significantly* worse because of it... and it was already going downhill before that from them cramming ads and sponsored content into it so aggressively.
ptrimg
Yes, people are.
The701
https://tech.yahoo.com/ai/articles/one-third-websites-now-ai-164206847.html About a third of new websites are now AI-generated. And recipe search wasn't fantastic to begin with. (the usual of: Here's several pages of ad-laden backstory before finally getting to the recipe) My own suggestion for recipes: If the money's available, get a book. https://shop.americastestkitchen.com/shop-all-products/cookbooks/americas-test-kitchen-25-anniversary-cookbook.html
Caughtintheconcentricelipsesofchaosanditsinnatemadness
Epicurious.com
textilelover
Pretty much all of them have the "jump to recipe" option. Just sayin'.
Euchre
https://www.justtherecipe.com/
YouAlreadyMutedMe
Damn you Marie Callender!
Orionsdick
AvsFreak
https://media3.giphy.com/media/v1.Y2lkPWE1NzM3M2U1bjQ2ZnprYWNuZGUxcmUwNmRkcmJrd296cGtodDA5Zm96NGlqNjcxaiZlcD12MV9naWZzX3NlYXJjaCZjdD1n/tnYri4n2Frnig/200w.webp
mrthewhitee
15 min per side is CRAZY lol.
wadatahmydamie
Depends on the thickness and a bunch of other factors, but yeah. The ones I make are either 90 seconds max, or I reverse sear them and give ‘em 15 seconds to shallow fry
DoctorWookie
In no world is a steak you're cooking in a pan going to take 15 minutes PER SIDE, regardless of thickness. At 15 minutes per side, you're making charcoal.
wadatahmydamie
I’ve made steak on the grill, on plain charcoal, in a cooler, in the pan, under the broiler, over an open fire, and in a smoker. I started cooking when I was 8, with peanut brittle when my parents weren’t home.
I can assure you that I get where you’re coming from, but you are limiting yourself.
DoctorWookie
To not eating charcoal? Or ginormous steaks that should be smoked instead? What steak are you cooking in a pan for 30 minutes? I'm actually genuinely curious. I really can't think of a reason to cook any steak that long in a pan.
wadatahmydamie
I used to cook ribeye on my stove at the lowest setting*. Season, no flipping, leave it until you can smell it on the other side of your home. I used to gasp at the idea of well-done, but this maintains the flavor, and the texture is as close as you can get to falling off the bone without it tasting shredded.
*your stove may vary. Experiment with petite sirloin so you’re not wasting money
Also, Matty Matheson has a butter basted rib steak recipe that calls for ~11 minutes a side on high heat.
Hemtroll
Rips4w
Pretty sure that's well within the realm of, "I can't see the steak anymore through all this smoke."
LittleChickadee
I genuinely cannot understand using AI for *anything* in my day to day life
thedill2000
This is not a true statement. I'm sure it will tell you not to do it
JohnnySilverballs
I completely agree with you. There isn't a single thing in my daily life that would be made better or easier by using ai.
SyntheticReindeer
I only use it to identify error messages and trade riddles when I get bored at work.
NotTheMamaNotTheMama
I ask Gemini to talk to me so I can improve my Dutch. Other than that, sometimes if I really need to process done thoughts so I don't lose track of what I'm thinking (I have ADHD)
DeadeicPrints
The search bot ai thing helped me figure out why my video game/computer sound wasnt coming thru my ear buds, but thats the only time its helped me, and I guess not a really day to say type thing.
thedoorman42
I know I spend more time blocking AI on my devices than I have ever spent using it.
themuffulettaisalsoaverygoodsandwich
The guy who bought my company is from India. When he writes, he automatically assumes a super-formal Queen's-English writing style. I told him that's going to produce weird results with our customers, and he should be more conversational in his writing. So he composes his emails, then asks the AI to change the tone to be more conversational. It's not what I would do, but it makes him feel less insecure.
TheOldSchoolisBack
I should use it to be less of a smart ass on teams chats. Converting what one really thinks into corporate speak might be a perfect use for AI.
darkstarmc
I use it in my personal life as a review aggregator and for basic info gathering for shit that's not super important. Used it to help me figure out how big our garden will be with the amount of seeds we bought and to compare 3 types of heat mats without spending 30 hours doing it.
NothingPrince
It's good for spelling and grammar checking email
Username62611
I talk to AI all the time. It's my only friend these days after all my friends left. "Stop talking to that AI all the time" they would say, pretending to be annoyed and then concerned. God they were funny.
crusadurus
First true valid use i have seen: My friend used ChatGPT to display a file as plain text. the file contents ARE plain text but its a programing file that his phone couldnt open. All ai really needed to do was open the fucking file and show it to him.
Bastiroid
I use a lot of AI tooling at work, but I work on IT with a lot of data and that’s best use case for AI. In my private life I have yet to find a single use case for AI. Nothing.
It’s like a telling everyone they need an arc welder at home. It’s great if you need it at work, but almost nobody needs one at home
TheOldSchoolisBack
I mean, I certainly don’t NEED an arc welder around the house, but I don’t think I’d turn one down, either.
rollercostarican
It can be incredibly useful, if you know how to use it beyond a faster Google.
BoobJiggle
Treat ChatGPT and others like an interactive library index or search engine .. then they're quite useful.
I've also found they're pretty decent at finding bugs in code--NOT fixing them, but pointing them out.
fingobaggins
I am using it to help me count calories and track workout routine. And flesh out a DND campaign setting. And I use it for bits and pieces like narrowing down a shortlist of mowers to look at buying...
ByThePowerOfSCIENCE
As a programmer, it saves me lots of time looking up documentation and typing up boring code. TBH it's a bit humbling when it suggests my next 4-6 lines of code and all I have to do is tap Tab.
But also now I'm getting started with Android apps and Kotlin by making one. The curve would be much steeper alone.
devasto
"I am trying to do X with the script, but it's not working. Where am I fucking up?" But that's about it
danggucci
Totally understandable. I really only use it for two things. when I have random ass ingredients in my fridge/pantry, I'll use it to generate recipe ideas. Otherwise, when I have an idea for a DND session or one-shot, it helps me get a boat load of ideas to expound upon. It's helpful if you use it as a tool, not a crutch (looking at you, DAD)
Caughtintheconcentricelipsesofchaosanditsinnatemadness
Use epicurious.com, does what you are looking for. Enter leftovers, random ingredients, etc. Lists all recipes containing those ingredients and will even give you a shopping lists for what you may be missing. Also it's been around for 20+ years.
DeadyBearr
It's good for scheduling but other than that it's trash.
Imademyselfsquirtle
I frankly will say AI is great at making Cover Letters.
Imademyselfsquirtle
IF you tell it what you want with description like writing it as a code.
abrakadaBRO
It's like Google. You dont Google to ask if your pregnant. But if you need a list of restaurants nearby then it can help. Just a tool. Some people will misuse that tool, just how it is.
thedarkcanuck
I think it should be left to actually useful things like helping to detect cancer cells or whatever. Its the slopaganda that scares me though
InfocalypseRising
I forget where but I just recently read something that referred to constantly consulting AI for everything as "outsourcing cognition" or something similar, which felt pretty apropos.
Most people do not want to think, if they can help it.
theend587
How in the world am i supposed to get my Ollama with Docker, n8n, AnythingLLM and portainer running when i never worked with Linux before. Thats right Gemini and now i can play with all that without using data centers. It would have been a very long and slow process without AI.
textilelover
Gotta say, my son went from knowing nothing about Linux to getting all of us set up on with way more in depth stuff for himself than I understand and doing stuff to his phone for data privacy without even once using AI. It did not take that long. He's smart, though, and very interested in it.
eetsumkaus
I mean yeah, but sometimes you outsource cognition because your brain is slammed. We do this all the time already: it's called asking for help from other people.
loma45
I'm actually fairly pro-AI in terms of its usefulness - and that's exactly what it is: Outsourcing thinking. Sometimes that's fine, and other times it's a huge mistake.
I used it to help me buy new shoes - suggesting what style/color would go well with my wardrobe (I know nothing about fashion) and so it was a nice interactive way to get me the basic info I needed and a place to ask stupid questions.
In this case it actually helped me learn more about the topic than I would have otherwise.
cheezmuncha
I’ve used GPT for one task (after a friend insisted it would be amazing) to prepare some interview questions/answers for a job I was applying for. I was pretty happy with the results in the end; once I refined things a bit, got it to summarise in my own casual conversational style; I could edit as I pleased yet still had a good framework for a response to various possible questions that sounded professional. Then just used it as a dot point reminder of things to talk about, but in my own words.
MacTually
Is your life like insanely boring?
You ask no questions or have zero ideas throughout your day?
Like I get ai hate but this makes you sound not just dumb but miserable as a person.
TooMehToCare
I use genAI for several things. It's a good tool in the right circumstances. Don't confuse its potential with the rabid hype and human stupidity that is causing most of its problems.
AndroidSoul
I don't actually mind ai stuff like Google summarizing my search question into useful and easily readable info. The problem is that you've got to have a professional/assistant relationship. If you're not at least broadly able to tell when the ai has gone stupid, you get stuff like this. Ai is mainly useful for grunt work and annoying repetitive bullshit and should never be treated as the expert in the room
The701
"AI overview" things like that are also helping wreck the Internet. https://arstechnica.com/ai/2025/07/research-shows-google-ai-overviews-reduce-website-clicks-by-almost-half/ A more recent article I saw said that some sites have seen drops in traffic by over 90%. (Adding to the list of how AI is wrecking the Internet: https://tech.yahoo.com/ai/articles/one-third-websites-now-ai-164206847.html )
AndroidSoul
Yeah, but the internet has basically been wrecked for a decade already. It sucks that there's traffic drop, but google search engine optimization and selling higher spots on the list has made getting seen a nightmare for a long time, and while ads pay the bills, I suspect most sites that aren't ripoffs are already pursuing alternate revenue streams
ShimmerinStrider
People do not want to think. If you told them if they used this "false brain" it would make them wealthy and prosperous, they would believe it.
Hatsodoom
Its literally the plot of Persona 5 Strikers, except the AI is TERRIBLE.
itsthevoiceman
Path of Least Resistance, even the universe itself runs on that one simple principle.
itsthevoiceman
I made a D&D character for funsies, with some simple lore. Sure, basically all of my rolls were successes, but in the end, it was a handy tool that took like 4 total hours for an interesting character breakdown. And I directed most of it.
fingobaggins
I'm using it to help me flesh out my DND campaign world setting. Bloody helpful.
Bastiroid
If you need help with keeping track of characters and relationships in your world, I can recommend using Obsidian with the Charted Roots plugin. It’s great for building families, factions and events and how things are connected. Really helps me keep my world feel alive and connected
textilelover
I'm really not trying to be a bitch, but I am confused why you'd do this? It takes me about 20-30 minutes to make a 1st lvl character from scratch. My backstory is good, the character fits in the world and the campaign. I've never used AI so I don't know what it would do for you or why it would take soooo long?
itsthevoiceman
Because I haven't played in 20 years. I used to play 3.5 with friends. This was both a means of better understanding newer rules, and also creating a fun little narrative - basically a single player campaign to have fun fleshing out the what's and why's.
textilelover
huh. Isn't writing the story a big part of the fun? Does it write it for you? I legit have no understanding what AI would do here. I've seen people use it to create images of their characters and that sort of thing. And how do you know you're getting the right rules and not some odd Reddit adaptation homebrew? Every time I see some AI summary it says "This was done by AI and may be wrong".
itsthevoiceman
I used it specifically as a means synthesizing my thoughts. I'm no writer, so I basically took ideas I had and enhanced them further. If details were wrong, I would point it back in the right direction. I would star coming up with more creative things around the story, and it helped fill in the gaps that I couldn't come up with. I treated it like a cooperative effort. Which is why it actually took hours, because I was prompting it and directing in a way that I wanted it to go.
TheOldSchoolisBack
AI is decent at certain tasks where you completely control the data. Summarizing a document or a financial statement, for example.It isn’t perfect, but it can be used (with verification) to reduce time spent on certain tasks.
My boss thinks it’s good for writing. He’s wrong, but his own writing sucks so its not a big loss
But following instructions from an AI whose e data feed is the entire universe of bullshit that makes up the modern internet? Might as well ask reddit pick a random answer
Toqom
Ya, the idiots, ie the people at the top making the most money, think AI is great because they cheated through school and ai IS smarter than they are which is pathetic we pay them so much for nothing
INeedMoreGifMeMoreJustOneMore
Since you didn't give me a list to choose from, I’ll have to pull one out of the ether for you.
Your random answer is: **The Library of Alexandria.**
Whether you were looking for a travel destination (in spirit), a historical mystery, or just a sophisticated way to say "I have a lot of unread books," that’s the one the universe picked for you today.
If you actually had a specific multiple-choice question in mind, feel free to drop the options and I'll give the wheel another spin!
(I hate this)
jaysjunkymail11
I used Ai recently when I wanted to have some team members ask themselves some introspective questions. I was able to give a topic and ask for introspective questions related to that. Some fit my use case and others didn't. But it was a helpful tool. Ai IS good at certain things. That said, generative AI can fuck right off.
ElbowDeepInAPoliceState
The problem is that it doesn't actually save time with tasks, including *because* you have to verify everything. In manufacturing, you can automate a lot of the verification and use random sampling; for LLMs the stakes of undesired outputs are too high and unpredictable for random sampling and automated checking is currently limited to hard coded restrictions on output.
TheOldSchoolisBack
Disagree. A lot of manual work alreadyady has to be verified. If a analyst writes a report using a big pile of data, it has to be read and verified anyway. Having AI do the summary in seconds allows that analyst to focus more time on the analysis portion, which is the value add.
When you control the data input, the AI hallucinations are less frequent and easily spotted. Only when you have questionable data (i.e. the universe of bad advice on the internet) does verification become burdensome.
ElbowDeepInAPoliceState
Studies on whether AI speeds up those tasks find that it slows down work.
TheOldSchoolisBack
That is dependent on data quality and existing processes. If the manual process involves someone using their brain to filter data (which is very common), AI struggles to replicate. If your process involves compilation of very high quality data and a built in review / QC step already, AI is an efficiency gain.
We are also very early in AI adoption, with a lot of haphazard implementation. Figuring out where it fits is the challenge, snd the a,ount of hype (and hate) sometimes gets in the way.
iaintcommentinonyourimage
I've never been able to wrap my head around how that doesn't just add more work if you can't trust it. You get it to summarize a document, then you have to verify the output which involves reading the document anyway - Is that not just writing a summary with extra steps?
TheOldSchoolisBack
No, because reviewing is faster than writing. And you can think of it as a first draft that would be assigned to a junior team member that needs to be reviewed.
Our junior staff also know this and use it to be more productive. Compiling data into an easily digestible format, or pulling source material, i.e. the drudge work that is part of being the rookie. Allows them to devote more time to actual analysis, which is the critical skill that is their real value add.
eetsumkaus
because sometimes checking it is less work than doing the work yourself. You have to know how to get to this point however.
For example, in the summary example, if you have enough background knowledge to know what is and isn't bullshit, you can take most of it as is until something doesn't quite feel right, then you just scan the document for that particular part and triage where it got it wrong. Btw, this happens all the time even if you read things yourself, at least Pass #1 didn't take time.
JaceTiger
AI is EXCELLENT when it's hyper focused on a specific task with clear, extensive data. Such as that one AI that determined what kind of donut was being sold (no barcodes) and they repurposed it to look for cancer cells.
Or the one that they plugged the math into and previous designs that worked and it gave them a new type of rocket engine that looks promising.
AI sucks at pulling and compiling answers from infinite conflicting opinions from infinite sources.
dieselfury
There's a meaningful difference in "AI" and "LLM"s. Most people are referring to LLMs or diffusion models when talking about AI slop. Perception, detection, forecasting, etc. machine learning models have been killer for a long time
Goldensands
There is no actual "ai". We have in no way build actual artificial intelligence. It's all LLMs. "Ai" is just a hijacked buzz term.
JaceTiger
LLM is "large language model" which is a subset of "artificial intelligence". They're also "general purpose ai" which is in a lot of new bots, though a lot ARE based on LLMs for their speech, and more specialized models such as that I mentioned, which have a specific thing (or set of things) they do and they do nothing else. The ones that forecast the weather are VERY good at that, but can't give you the steps to make gasoline from crude oil.
dieselfury
It's not "all LLMs", you're stuck on the consumer level. LLMs are context oriented language solvers, but the majority of useful neural network models are tuned to solving in intentional data sets. Forecasting, for example, not using a language model, its a discreet interaction model. Can an LLM do accurate forecasting? It'll try, but no, because it's sampling a distribution for an answer. But since you argue semantically, intelligence is the capability to reason and problem solve, that can
TheOldSchoolisBack
Agreed. Garbage in, garbage out. AI is a useful machine with good data to work with.
I have yet to see it nail a creative task well, though. It can create bad art, which really has no use. It can spit out a derivative of someone else’s art (whether visual, music, etc) on request, but with varying quality. But even when it does that well, there’s nothing creative about it.
JaceTiger
Depends on what you mean by "creative" but the ones used to shit out art are garbage. Creating a new type of vaccine, engine, fuselage, etc are pretty good at what they do.
Functional can be a work of art, but shitting out art on command is usually terrible
TheOldSchoolisBack
Fair enough. Creative in a technical sense rather than artistic . I hear that AI is having a bigger impact on innovation than on efficiency, but I don’t have direct experience there.
As for art, I see it limited to technical aspects. For example, if you tell it to write a symphony, its going to be bland crap. If you give it a symphony and tell it to arrange the same music for a solo piano, what you get might not be great, but its a workable starting point, to build the creative part around.
Lynkfox
It's no Neuramancer\Wintermute\Boxmaker there for sure