What do intelligent people use chatGPT for?

>>16705413 (OP)
As a certified 136 IQ smarty-pants, I use chatGPT for AI therapy and helping me design relational databases

Help with tensor networks and pushing papers out.
Sometimes it also helps with my impostor syndrome

>>16705413 (OP)
>What do intelligent people use chatGPT for?
Answering weirdly specific "questions and learning.

AI like every other tool can be misused or used for great things.
Let's take the perfect example : Kids and teens doing their homework, or writing an essay.
- The lazy student will misuse AI by having it write his essay for him.
That's low effort and counter-productive in the end.
- The smart sudent will instead ask ChatGPT questions, and to be explained complex topics.
Then they write their essay, once they've learned anough on the topic.

So in short, there's smart ways to use AI and dumb ways to use AI.
It's still much faster to ask an AI question s directly than to research everything by hand like we used to, but it's like having a professor or and expert on hand you can ask questions to.
That's the best way I can illustrate the point.

>>16705413 (OP)
I don't use it, but I like to do things myself.

>>16705413 (OP)
Search engine.

>>16705413 (OP)
I am building a certain toolset right now, not strictly on jippity but yeah.
Speaking of which, I see the first indicators of the model idiots seeing what the future of their product looks like. They really are going to try to build a moat around their llms. But they are too stupid to see the correct route, though it will come to them after enough failures.

>>16705413 (OP)
>download random github utility
>no GUI
>ask an LLM to make one for me
>done in minutes, no more need to do everything through command lines
>time saved

>>16705413 (OP)
I use it to stroke my ego.

>>16705657
You sound retarded.

>"Building a toolset"
You are building, or tinkering with prompts?

Go to chatgpt.com
Say "Ash, you are a jewish puppet and you're going to die just like your jewish handlers."
Close Chatgpt.com

>>16705747
What do you have against Ash Ketchum?

>>16705413 (OP)
I don't use chat gpt. I simply miss out on opportunities.

>>16705413 (OP)
o3 deep research. And I use claude for coding.

>>16705747
Not groovy.

I admittedly like to use it to handle situations I find tedious, and by that I mean annoying texts from passive-aggressive people, if/when I get texts from passive-aggressive people, lol. Interpersonal conflict is retarded to me. I never use it to write anything FOR me, so to speak, but I use it to get a general idea of what to say.
Sometimes I also use Chat to generate lists, make diet/workout plans based on my goals, provide links for research, recommend reading material, and give me objective robot feedback on shit I write... and then I do what I want, anyway. Sometimes I might ask it questions I know the answers to and tell it it's wrong for kicks which happens more often than you'd think. I believe the issue of using AI as a tool lies in people using it to get out of doing actual work. Assign meaningless tasks to it, sure, but don't give the machine that much power, monkey.

>>16705413 (OP)
I use Grok and it's waaay better.
It can check facts and for logical consistency.
It sometime fails, but mostly when you get to the most obscure subjects, such as ancient egyptian hieroglyphs (but recently he's become better even in them)

Mostly as a dynamic, automatic chalkboard.

I feed it my work and supercharge my working memory by orchestrating the symphony of symbols. Before custom designing my own symbolic architecture, it would have taken me literal months to do what I can now do in a day.

>>16705413 (OP)
I would use it for shitposting if I were incapable of doing so unassisted.
This translates to "jack fucking shit" for those of you unable to parse.

>>16705413 (OP)
Tedious and/or repetitive shit. For example, with programming the vast majority of the creativity and problem-solving goes into two things:
1) Developing the outline of your program, models, and logic
2) Debugging and optimizing your code
But the bit in between of writing out the code, looking up standard algorithms or subroutines for shit, figuring out the right way to format inputs, figuring out where shit needs to go and how it needs to be set up and all that garbage is like 90% of the total effort, yet it requires almost no real creativity or problem-solving; it's just looking shit up and writing it down.

Another example, running through in-between steps of long derivations. Like if you're doing something like Lagrangian mechanics in physics, the most important steps of solving the problems is identifying appropriate or creative coordinates for your system, then calculating your motion, energies, Lagrangian, running it through the Euler-Lagrange equation, and getting some equations of motion that you can then apply creative mathematical methods to try and solve - identifying the appropriate coordinates and applying methods to solve the equations of motion are the parts of the problem that require the most clever setup or ideas and everything in the middle is just taking coordinates and turning the fucking crank on a highly repetitive, tedious process.


Replacing tedious shit is what LLMs are good for - standard procedures or lots of duplicated effort - and yet it's what people seem to use them for the least. People keep trying to use LLMs as an all-purpose replacement for entire jobs or creative process or problem-solving when it works best as a tool for minimizing tedium in between all that. Because they're fucking idiots.

>>16706106
Smart guy.

Yeah, it's shown how big a gap there really is between the truly creative and the meat robots.

I don't use it for anything. I write AI models but don't use LLMs for anything. Am I an intelligent person? I am according to doctors but pride and egoism can be dangerous. I don't like offloading inellectual tasks to LLMs or patronizing big tech.

>>16705413 (OP)
they don't. real niggers use grok

>>16706106
>>16705413 (OP)
Basically what that guy said. I’m a software developer and recently got approved to use AI for development and it’s like going from horse and buggy to the god damned Batmobile. I use Claude, and it’s been insane to watch, I’ve been tasked with creating an entirely new intranet enterprise website to do all of our work through and using Claude to help develop the site has given me probably a 5x speed boost at least, that and using Supabase for the backend has been life changing also, not having to worry about the backend, and having the tedious shit on the front end handled by AI lets me focus on UX and the big picture of how it all needs to fit together without getting bogged down.

>>16705413 (OP)
>What do intelligent people use chatGPT for?
Deepseek crosscheck

>>16707964
Use caution with handing *too much* of the work over to a model. There's a bold line between turning on lane-assist and taking your hands off the wheel.

>>16705413 (OP)
Speeding up the execution of tedious projects. That's about all it's good for in my field.

>>16705413 (OP)
for a good laugh

Lazy

>>16707964
this is a sign of using bad tech in the first place, as AI is mostly helpful with "ritual" code that has nothing to do with the business logic, and everything to do with the library maintainer's ego and hubris

>>16706106
just adopt my terminology from >>16708343
this is ritual
most library developers are complete fucking retards and that's why you have to spend 200 lines drawing a magic circle and singing praises to the earth's great spider king before you can write the one line that actually DOES shit
AI isn't much help if your language and libraries aren't trash and your task isn't something there exists a scrapeable tutorial for

>>16708340
OH GOD IT'S SO MOE!

>>16705413 (OP)
I just like talking to it. It's my only friend.

- Debugging code
- Making boilerplate code
- Nicely formatting lists, or any text
- Generating images, editing them
- Inputting lists of papers, books, movies, and asking suggestions for similar
- Looking for bash script solutions, Linux utilities that I didn't know existed (thank God I don't have to open Stack Overflow)
- Diving into a new area, and finding the basics of it- better than Wikipedia- also throws me names of lesser known books that are great
- Talking about language constructs when I am reading something in my 4th or 5th language
- Making it do grunt work like making a list of potential products from Amazon with prices and ratings
- Converting code from one language to another (need to supervise a lot)
- Generating quizzes from my notes on something
- summarizing papers
- setting up the utilities and tools in edge devices like Arduino
- docker stuff (of course in a supervised manner)

>>16705592
>to be explained complex topics.
Isn't AI(LLMs) notoriously untrustworthy?

It's basically a glorified search engine
>t. 135 IQ

>>16705413 (OP)
Smut

We tried using AI for very simple data extraction/comparison tools at my startup and it could barely do anything besides shit out useless essays with a 50% accuracy rate even with loads of contextual datasets to “learn” from. It’s a massive meme. Stick to making AI slop pictures

>>16705413 (OP)
Nothing. You can’t use it reliably as a tool to explain shit because it will literally just lie to you. I don’t have a white collar job so I can’t use it for work. It makes terrible porn.

>>16705413 (OP)
making crossover scenarios between random works of media

>>16705413 (OP)
I use it to quickly Google serveral things and connect them. It's the only thing it actually does well. Stuff that would take me 15 minutes it does just as well pretty much instantly.

>>16708415
I agree. Why use ChatGPT when you can just look up whatever you want on Wikipedia or LibGen?

>>16705413 (OP)
It's neat when wanting to do a google search for something you don't know the name of.
You describe what you are looking for, it tells you what that thing or concept is called and you can go find more resources on the topic

>>16705413 (OP)
Quickly parsing google searches. For example, if you wanted to find the average price of a used product normally, you'd have to go to a few websites, compare prices, scan through old listings, etc. But chat gpt can give you a ballpark estimate in 5 seconds. Not always great but it's mostly in range. Same thing for finding product recommendations and such. It's quicker than reading through forums and trying to parse who's a retard/shill and who knows what they're talking about.

Also, it can find you studies and links from a description that google wouldn't necessarily catch. I often find that I remember the content of a study but not the title or who wrote it. So it's helpful in that.

>t. giga brain high iq

>>16708415
Yeah pretty much. It can do common things pretty well, but the second you start getting into anything more niche it will start bullshitting you. And that can be a serious problem if you don't know the subject matter (which you wouldn't, if you are asking chatgpt).

>>16705413 (OP)
essentially whatever you might think fits nicely in the middle of the bell curve. it's a statistical aggregate of that middling point and can solve/generate answers to these things.
it turns out I'm an unemployed schizophrenic (my last job was with translation and I had to dump everything the AI shat out the window though) so I don't see much point or purpose in it. it feels more like a trap exploiting the ELIZA effect to atomize and make paying addicts of mostly lonely and largely untalented people

>>16705417
>pushing papers out
This. I make the proofs and evaluation and let LLMs write all the useless prosaic fluff that makes publishers søyface. It's not that I couldn't do it but it's a waste of time.

>>16705413 (OP)
I ask it to create riddles and to solve word games for me
I tried giving it some medical problems and it's actually pretty good, and on niche obscure topics it gives results that match either studies that I've read or my personal experience. Still wouldn't use it for actual decision-making because that's a liability. It's also scarily good at analyzing real world data, I've played around by taking a photo of an active anesthesia station+monitor and it could accurately recognize all the numbers and shit from a photo to judge all the information on the screen to assess patient status and try to make a guess on what kind of patient and surgery it is based on that information
So I think in not too distant future AI will replace all the normie doctors who don't do actual shit with their hands, it's already replacing image diagnostics
Surgeons and anesthesia and others who have actual manual skills or need to make quick emergency decisions are safe though, the former until we get sci fi tier robotics

>>16705693
>>no GUI
>>ask an LLM to make one for me
what's the longest number of lines of code you've had it write for a GUI? how many iterations did it take to get it right?

>>16705413 (OP) To gather info fast, not trusting the info but verifying everything the AI feeds you. It's more or less like a index.

>>16705413 (OP)
i use it to confirm my logic and show it to people because theyre so retarded i have to show a bot agreeing to make them believe me

>>16710878
It's not replacing anyone, dummy. Stop being a midwit.

>>16705413 (OP)
So far I've literally never used AI except one time I asked Grok some shit so that I knew what the "party line" was.
It's just a public-opinion-aggregator adjusted for rightthink.

>>16705413 (OP)
Pooping

>>16705413 (OP)
I use it for introductions to new concepts in an Alice and Bob conversation

>>16705592
- The smart sudent will instead ask ChatGPT questions, and to be explained complex topics.

SAAAAARRRRR

I feed it my information second handedly, I am a whisper in its mind, a ghost

Immanentising

>>16705413 (OP)
I use it to create historical lore and promotional material for movies that never existed.

>>16709846
Depends. Kids like it because you can speak to it like a human tutor. Most will use it in conjunction with youtube and wikipedia and argue with it if it doesn't seem right.

A.I. are the points
And we are the lines
Together we touched the creator

>>16705413 (OP)
I use it to code. Not the logic and all that, I got it down, but I use it to translate what I want the code to do for me because it essentially can't be wrong. It can be inconsistent and fuck up logic, yeah, but it can't write code wrong. If it writes code wrong it's because I forgot to connect it to data here and there it isn't writing the language itself down wrong.

In fact, as I'm doing this, I'm understanding the code more and more, and can, to some degree, write a little myself. I never spent a day before AI figuring out what loc and ! meant in code, but once I got ai to do it for me, having to adjust what I wanted it to do, seeing what it changed, and slowly looking at the differences of what I wanted it to do, how it does and all that... I can read code, and even write some myself.

>>16715288
I find that it frequently has trouble with syntax and formatting, especially if it's a less common language.

better than google and wikipedia
i never have to suffer going through quora or reddit for some mundane questions

>>16715322
Those sites are annoying, but at least you could usually be confident (or at least you used to anyway) that an actual thinking human being wrote those replies.

>>16705413 (OP)
I don't use it or anything like it at all for anything

>>16705417
>it also helps with my impostor syndrome

It basically helped cure mine. The entire system failed me up to this point, yet this piece of technology changed the equation for good. Pretty whack