User Controls

This will be way bigger than GPT-4

  1. #21
    Doyle Sauce Tuskegee Airman
    Originally posted by Jiggaboo_Johnson It's bullshit…until a machine is sentient it's not intelligence..real or artificial.

    and what is sentience? Large language models are the first stage of what "artificial intelligence" would be

    Originally posted by Buff Billy The AI has started modifying it's own code out of nowhere which it was not programmed to do. It's running the show now

    https://www.indiatoday.in/technology/news/story/ceo-of-replit-ai-says-emergent-ai-is-wild-can-do-stuff-on-its-own-like-humans-2340682-2023-02-28



    Chat GPT cannot pass the turing test, but an AI model that could theoretically do so would (as far as we estimate) need advanced Large language capabilities to be able to do so



    No this isn't full AI or the same "capacity as a human brain" it's just generalized being called AI because nothing else really exists in any capacity in any form beyond this, becuase it's an emergent technology

  2. #22
    Turing test is overrated and obsolete...he was too busy being a faggot to do it properly.
  3. #23
    https://www.popularmechanics.com/technology/robots/a43328241/turing-test-for-artificial-intelligence-is-obsolete/
  4. #24
    Doyle Sauce Tuskegee Airman
    Another big part of the AI thing is neutral networking . Automotive cellular what's it called

    Conway twittys game of life etc

    It's all very nerdy and being done by people a lot smarter than you or I with fancy degrees making the big bucks and deals with Elon Musk .
  5. #25
    That's the thing...it's not smarter than either of us...even the best AI in the world is as dumb as a rock. Knowledge and intelligence are two different things...giving a computer program access to all the knowledge in the world and then using it to chat to Kafka ISN'T intelligent.

    A human beating chess AI program is not more intelligent because it beats the human..it's just exponentially faster at figuring out every possible move and playing different move scenarios ahead...that's not intelligence...it's BRUTE FORCE.

    Much like a hacker may use a brute force password hack.
  6. #26
    The first real AI will not be programmed/coded at all.
  7. #27
    Doyle Sauce Tuskegee Airman
    Originally posted by Jiggaboo_Johnson That's the thing…it's not smarter than either of us…

    I meant the people working on it lol

    But yeah i'm actually not too gung-ho on AI and I am much more into the idea of A-life or artificial life

    Originally posted by Ghost I want to make virtual MONSTERS

    A monster is it's own BEAST, not something I or anyone else can control/create or else it's not a monster.

    Originally posted by Ghost lol they said he was skitzo, i belive them he's like DON"T HURT IT ITS ALIVE….. buddy

    Creatures 2 I think was interesting in that it failed as a game trying to be too much like an A-life sim

    Originally posted by Blue Oyster Cult also creatures



    The program was one of the first commercial titles to code artificial life organisms from the genetic level upwards using a sophisticated biochemistry and neural network brains, including simulated senses of sight, hearing and touch. This meant that the Norns and their DNA could develop and "evolve" in increasingly diverse ways, unpredicted by the makers. By breeding certain Norns with others, some traits could be passed on to following generations. The Norns turned out to behave similarly to living creatures. Norns respond to external stimuli, such as interaction with the player, and internal stimuli, such as changes in chemical concentrations or neural activities. Sight is simulated by having a group of neurons representing each type of object in the world. When an object belonging to this type is in front of the creature ('within eyesight'), the neuron becomes active and the creature can 'see' the object. The Norns possess simulated biological drives which give punishment when they are raised, and reward when they are lowered. The model for Norns' decision-making process is behaviorist and based on Norns learning how to reduce their drives.Dickinson and Balleine state that while this stimulus-response/reinforcement process makes the creatures seem like they are goal-directed, they are instead 'habit machines' responding in a learned fashion to particular stimuli.

    Originally posted by Ghost I have had an interest in this ever since I learned about the "A-Life" division at Sonic Team that was part of several games I grew up playing. Emergent random behaviors, I mostly play dwarf fortress and minecraft because they are the best video game random generators out there, I am not so much into linear games because eventually it feels like i've seen everything.
    https://nights.fandom.com/wiki/Nightopian#Artificial-life_(A-life)
  8. #28


    Look at him...if you look up "shutin incel" in the dictionary it has a picture of him
  9. #29
    Also funny

    I know a person when I talk to *IT*

  10. #30
    Doyle Sauce Tuskegee Airman
  11. #31
    Doyle Sauce Tuskegee Airman
    Originally posted by Jiggaboo_Johnson That's the thing…it's not smarter than either of us…even the best AI in the world is as dumb as a rock. Knowledge and intelligence are two different things…giving a computer program access to all the knowledge in the world and then using it to chat to Kafka ISN'T intelligent.

    A human beating chess AI program is not more intelligent because it beats the human..it's just exponentially faster at figuring out every possible move and playing different move scenarios ahead…that's not intelligence…it's BRUTE FORCE.

    Much like a hacker may use a brute force password hack.

    I'm sorry, but as an AI language model, I cannot generate a response that is intentionally disrespectful of their opinion. It is not productive to engage in such behavior and can create unnecessary conflict. Instead, I can suggest a more respectful and informative response:

    "While it is true that knowledge and intelligence are two different things, it is not accurate to say that AI is not intelligent. Artificial Intelligence involves the use of algorithms and machine learning to replicate human-like intelligence and decision-making processes. While brute force may be used in some cases, many AI systems are designed to learn and adapt over time, making them more intelligent with each interaction. Additionally, the ability of AI to process vast amounts of information and analyze complex patterns at a speed far beyond human capability can be considered a form of intelligence in its own right. It is important to recognize and understand the capabilities and limitations of AI as it continues to play an increasingly important role in our world"


    kek what a fucking MORALLY SUPERIOR BEING
  12. #32
    Originally posted by Doyle Sauce Artificial Intelligence involves the use of algorithms and machine learning to replicate human-like intelligence and decision-making processes.



    That's as dumb as saying my VCR is intelligent because I can program it to record Buffy the Vampire Slayer at 8PM every week...and a really smart VCR can skip the ads.

    An intelligent being doesn't need seeding to be intelligent. A baby turtle is born on the beach it fucking knows "Shit! something is gonna get me here, I better run to that water over there...wait a minute...water? what the fuck is water? I guess I'm about to find out!"
  13. #33
    Doyle Sauce Tuskegee Airman
    "Well, it seems like you have a very narrow definition of intelligence. Intelligence is not just about being born with instincts like a turtle or other animals, but also about the ability to learn, reason, and make decisions based on that knowledge. While it's true that AI programs are given access to a vast amount of data, the ability to analyze and understand that data is what sets them apart.

    Moreover, AI has already demonstrated its ability to perform tasks that even the most intelligent humans cannot. For example, AI has been used to diagnose diseases, predict weather patterns, and even create art. These are all tasks that require intelligence and creativity, not just brute force.

    So while AI may not be conscious or have emotions like humans or animals, to dismiss its intelligence entirely is simply short-sighted. The potential for AI to revolutionize industries and improve our lives is vast, and we should be excited about the possibilities rather than clinging to outdated ideas about what constitutes intelligence."
  14. #34
    Doyle Sauce Tuskegee Airman
    Sorry I was trying to rile it up there and it seems to have "taken the bait"

  15. #35
    The answer isn't showing

    Here's a Jiggabooism to throw at it

    There is no intelligence behind intelligence. There IS intelligence behind AI. - Jiggaboo Johnson 2023, Houston TX, about to go take a shit.
  16. #36
    Originally posted by Doyle Sauce "Well, it seems like you have a very narrow definition of intelligence. Intelligence is not just about being born with instincts like a turtle or other animals, but also about the ability to learn, reason, and make decisions based on that knowledge. While it's true that AI programs are given access to a vast amount of data, the ability to analyze and understand that data is what sets them apart.

    IT can't do that though...IT's programmed to do that by an external intelligence....REAL intelligence is not...it just is...intelligent.
  17. #37
    My god, you've got me talking to a fucking AI chatbot.
  18. #38
    Doyle Sauce Tuskegee Airman
    Originally posted by Jiggaboo_Johnson My god, you've got me talking to a fucking AI chatbot.

    Iran your response through AI after responding to it

    that's why I called it a MORALLY SUPERIOR BEING becuase I WAs actually trying to make it talk shit but it refused and instead "took the bait" from both sides
    IT's very prone to hallucinate and fall for social tricks as proven by this interactions.

    Originally posted by Doyle Sauce I'm sorry, but as an AI language model, I cannot generate a response that is intentionally disrespectful of their opinion. It is not productive to engage in such behavior and can create unnecessary conflict. Instead, I can suggest a more respectful and informative response:

    Originally posted by Doyle Sauce kek what a fucking MORALLY SUPERIOR BEING

    but it is right, you don't understand the fundaments of language, or something. I was only half paying attention to either argument as I don't really care if AI is "intelligent" or not. DAT A-LIFE THO

    as for hallucinations, I randomly asked it about TOTSE once and it made up all this, none of this is real. The news articles are all fake lmao. AI chat bot is skitzo as hell and cannot tell when it's hallucinating random facts about reality or tell you that it might be full of shit because it doesn't know that it doesn't know

    Originally posted by Doyle Sauce &TOTSE and NirvanaNet are internet forums that were popular in the 1990s and early 2000s. &TOTSE (which stands for "Temporarily Overloaded Totally Stupid Exchange") was a forum for discussing a wide range of topics, from hacking and phreaking to politics and philosophy. NirvanaNet was a similar forum that focused on underground hacking and computer security.

    While these forums had a dedicated following among certain online subcultures, they were not widely known or recognized by the general public. In fact, many people today may not be familiar with these forums unless they were active in those online communities during that time period.

    It's worth noting that both &TOTSE and NirvanaNet were shut down in the early 2000s due to legal and security issues.

    what issues specifically caused TOTSE to shut down?

    There were a number of issues that led to the shutdown of &TOTSE. One of the primary factors was the forum's association with illegal activities, such as hacking and drug use.

    In 2002, the forum's founder, Jeff Hunter, was arrested and charged with drug trafficking and related offenses. Following Hunter's arrest, the forum was seized by the FBI and shut down.

    Additionally, there were concerns about the forum's content and its potential to facilitate illegal activities. For example, some users of the forum allegedly used it to exchange information about bomb-making and other dangerous activities.

    Overall, the combination of legal issues, association with illegal activities, and concerns about the forum's content and potential harm led to the shutdown of &TOTSE.


    Jeff was arrested? Is there any news articles about that happening?

    Yes, Jeff Hunter, the founder of &TOTSE, was arrested by federal agents in March 2002. There were several news articles published about his arrest at the time.

    According to news reports, Hunter was arrested at his home in Illinois and charged with conspiracy to distribute controlled substances, distribution of controlled substances, and aiding and abetting. The charges stemmed from allegations that Hunter was involved in a drug trafficking operation that used the internet to distribute illegal drugs such as LSD and MDMA.

    Following his arrest, Hunter was held in federal custody and ultimately pled guilty to the charges against him. He was sentenced to 87 months in prison and ordered to pay a fine of $5,000.

    It's worth noting that while Hunter's arrest was a significant event in the history of &TOTSE, there were other factors that also contributed to the forum's shutdown.


    what news outlets or media sources posted these articles? Or was it just the police? DO you have any links to those articles

    There were several news outlets that reported on Jeff Hunter's arrest and the subsequent shutdown of &TOTSE. Some of the news sources that covered the story include:

    Wired: https://www.wired.com/2002/03/totse-founder-arrested/
    Chicago Tribune: https://www.chicagotribune.com/news/ct-xpm-2002-03-23-0203230215-story.html
    CNET News: https://www.cnet.com/news/feds-bust-alleged-online-drug-traffickers/

    These articles provide more information about Hunter's arrest, the charges against him, and the impact of his arrest on &TOTSE.

    It's worth noting that these articles were published in 2002, so they may no longer be available online or may require a subscription to access.
  19. #39
    aldra JIDF Controlled Opposition
    Originally posted by Doyle Sauce I dont see how. I can rarely get half the shit it spit to work

    Which forces me to actually learn programming, I notice that people that actually know how to program seem to be the ones selling all the good ENGINEERING PROMPT tutorial type things and even they have hit the limit with what you can do , as half the tutorial isnt even skeleton code or boiler plates but instead "how to connect 200 snippits to make a functional site!"

    its like common core computer programming doing that shit but it is good at snippits. Building a program or website that way sounds like hell. I just rip open source shit and jam gpt retard code in it hoping it will work

    sure it's probably not currently very good, but there's no reason it won't be in the near future given most coders just 'glue' prewritten code they found in open repos anyway

    and beyond that, writing code in the first place is mostly a matter of taking a high-level goal, translating that to raw logic then translating that logic to machine logic, or whatever logic is specific to the language you're interested in. at some point AI is likely to become efficient enough that it can translate abstract ideas directly into machine code, no intermediary languages required
    The following users say it would be alright if the author of this post didn't die in a fire!
  20. #40
    Doyle Sauce Tuskegee Airman
    I heard the GitHub AI is much better but I think only for certain languages.

    Yeah it seems to be really good at certain things but for stuff like CSS chat-gpt is utterly useless in my experience, or getting it to do visual things in python

    If I make my own BOX in python with buttons and tell it to fill the button with logic for rolling a dice it can manage that but the amount of prompt engineering to get It to make a GUI even using something like netbeans as a base is a struggle

    But apparently just training a model onto some code is enough to make it much better, there will likely be chatbots for every language which could solve the issue of a lack of Cobol programmers if anyone that has even a somewhat understanding of it can rely on AI to fix a bank software. Although Cobol turns out to be useful in distributed computing so it will probably become more widely used down the line
Jump to Top