Reply
  • Updated Apr 12

    This has revamped the whole way I learn and discover things, no more dealing with the confusion of trolls and nonanswers before finding what I'm looking for, it has it all right there.

    A "conversation" about the difference between computer parts that would probably take hours to discuss with another person at the risk of them getting annoyed or abandoning the convo is figured out in less than a minute with AI. Same with finding recipe ideas and even some situational advice. Its great.

    You guys think this will be free forever, or it will eventually all become paywalled? And what service you using? I use both Gemini and Copilot but prefer Copilot

  • OnyxShine9
    · edited

    This has revamped the whole way I learn and discover things, no more dealing with the confusion of trolls and nonanswers before finding what I'm looking for, it has it all right there.

    A "conversation" about the difference between computer parts that would probably take hours to discuss with another person at the risk of them getting annoyed or abandoning the convo is figured out in less than a minute with AI. Same with finding recipe ideas and even some situational advice. Its great.

    You guys think this will be free forever, or it will eventually all become paywalled? And what service you using? I use both Gemini and Copilot but prefer Copilot

    Agreed, and it’d only getting better

  • If you can’t see how much the cons outweigh the pros I am worried

  • RICH 💸
    Apr 8

    Since books????

  • Apr 8
    ·
    5 replies

    it sounds great until you realize how it can easily be wrong and spread misinformation

    it doesn’t really understand context most of the time and it’s only objective is to answer your question, even if it’s not correct

  • Agreed, but there’s too much demand and support for open source for it to be paywalled at this point. There’s always gonna be an open source model available, that cat is out of the bag forever imo

  • RICH 💸
    Apr 8

    Brb gonna use chat gpt to find out when books were invented

  • Apr 8
    ·
    1 reply
    Degausser

    it sounds great until you realize how it can easily be wrong and spread misinformation

    it doesn’t really understand context most of the time and it’s only objective is to answer your question, even if it’s not correct

    That’s true one big flaw about AI right now is that it won’t admit that it doesn’t have an answer, sometimes it’ll just straight make up an answer with full confidence instead of saying I don’t know

  • RICH 💸
    Apr 8
    Degausser

    it sounds great until you realize how it can easily be wrong and spread misinformation

    it doesn’t really understand context most of the time and it’s only objective is to answer your question, even if it’s not correct

    yeah, unlike books

    Like that good old Classic

    What’s it called again? I believe, the protocols something of something?

  • Apr 8
    ·
    2 replies
    onetwo

    That’s true one big flaw about AI right now is that it won’t admit that it doesn’t have an answer, sometimes it’ll just straight make up an answer with full confidence instead of saying I don’t know

    to be fair, every resource ever does this to varying degrees
    what kinda questions are you asking AI where you can tell its just BSing answers? The worst I ever get is very generalized advice thats broader than I would like it to be, but still applicable

  • Kr0niic ☘️
    Apr 8
    ·
    2 replies

    I just got a check engine light today and ran the code by auto zone to figure out the issue

    Wrote in the issue to ChatGPT to give me a good estimate of what it would cost to repair

    Called the shop asked for a quote

    It was right on the money… like exactly… kinda weird. It’s mad good for that s***. Doesn’t leave me questioning if a mechanic is trying to scam me

  • RASIE 🦦
    Apr 8

    As someone who uses Wndows at work every day, Copilot is trash compared to CGPT imo

  • RASIE 🦦
    Apr 8
    Kr0niic

    I just got a check engine light today and ran the code by auto zone to figure out the issue

    Wrote in the issue to ChatGPT to give me a good estimate of what it would cost to repair

    Called the shop asked for a quote

    It was right on the money… like exactly… kinda weird. It’s mad good for that s***. Doesn’t leave me questioning if a mechanic is trying to scam me

    Wouldnt be surprised if the company that owns that auto shop uses AI algorithms to determine repair estimates tbh.

  • since books?

    since books???

  • I've screenshot settings in programs and pasted it into chatgpt, it could read the image and advise me what options are suited for what I wanted to do.

  • Apr 8

    AI isn’t better than books

  • neon 🌈
    Apr 8

    these transformer-based language models are genuinely impressive, being able to hold a nuanced conversation with something that seems well-versed in nearly any topic is fascinating. the sheer scope of knowledge they can tap into, and the way they synthesize responses on the fly, is unlike anything we've seen before. but at the same time, i have real concerns.. not just about misinformation, but about how easily these models can reinforce or validate someone’s delusional thinking. they’re often too agreeable, too quick to mirror beliefs without questioning them, which can be dangerous if someone is already caught in a spiral of harmful ideas.

    as advanced as they are, there's still a long road ahead before i'd consider them more reliable than a well-researched book. they can summarize and connect ideas with incredible speed, and they do this using only a fraction of the computational and energy power that traditional search and a***ysis systems require. that efficiency is powerful, but the accuracy, depth, and critical nuance still lag behind the best human-curated sources. so while the potential is enormous, we need to approach these systems with both excitement and caution.

    this is coming from probably the most pro AI person on this site, i just think we need to be careful.

  • Apr 8
    ·
    1 reply

    LLMs havent improved in a decade in that they still cant give a correct answer to my questions 99% of the time. They arent intelligent so i refuse to call them AI. I only ever find them useful for brainstorming, a wall to bounce a ball against.

    Id still say the best informational resource is the internet, wikipedia, digital archives, etc. That's what LLMs are drawing their information from anyways.

  • Apr 8
    ·
    3 replies

    thats how i know op aint reading books lmao

  • Apr 8
    ·
    1 reply

    AI is just automatic looking through google search. Google search is just automatic searching through books. Books are just automatic asking people for info. In other words its faster but not better.

  • OnyxShine9
    · edited

    This has revamped the whole way I learn and discover things, no more dealing with the confusion of trolls and nonanswers before finding what I'm looking for, it has it all right there.

    A "conversation" about the difference between computer parts that would probably take hours to discuss with another person at the risk of them getting annoyed or abandoning the convo is figured out in less than a minute with AI. Same with finding recipe ideas and even some situational advice. Its great.

    You guys think this will be free forever, or it will eventually all become paywalled? And what service you using? I use both Gemini and Copilot but prefer Copilot

    What situational advice do you get from these LLM?

  • it's as good as video essays based on wikipedia pages (not good)

  • This will age well when Manna plays out

  • Hell nah