(Topic ID: 337395)

Using ChatGPT to repair pins

By OzyPinball

11 months ago


Topic Heartbeat

Topic Stats

  • 103 posts
  • 43 Pinsiders participating
  • Latest reply 4 months ago by arcyallen
  • Topic is favorited by 2 Pinsiders

You

Linked Games

No games have been linked to this topic.

    Topic Gallery

    View topic image gallery

    Screenshot_20230528-161245~2 (resized).png
    finally-the-mystery-of-the-three-seashells-has-been-solved-v0-mkscv1k80rga1 (resized).png
    pasted_image (resized).png
    IMG_4495 (resized).jpeg
    IMG_0083 (resized).jpeg
    IMG_3640 (resized).jpeg
    chat2 (resized).png
    Chat1 (resized).png
    There are 103 posts in this topic. You are on page 1 of 3.
    #1 11 months ago

    I'm having problems getting a Lord of the Rings to update ROMs correctly, but can't find anyone who knows what's going wrong. I turned to ChatGPT in desperation. It surprisingly had some detailed information about what I was working on but ultimately wasn't very helpful. Some of the details were wrong (like the ROM version) and it was missing other important steps.

    There is a vast wealth of technical knowledge on Pinside... It would be great if it could be dumped into a chatbot some day, able to come up with novel solutions to unique problems.

    Has anyone successfully used chatgpt to actually diagnose and solve a problem?

    Chat1 (resized).pngChat1 (resized).png

    chat2 (resized).pngchat2 (resized).png

    22
    #2 11 months ago

    Stick with Pinside and Google.

    Too much A I is going to wreak us humans.

    26
    #3 11 months ago

    ChatGPT, and any other AI tool, is no where near 100% accurate. Don't risk ruining a game by trusting a computer.

    These "AI" chatbots are glorified Google search engines being fed by thousands of data sources.

    #4 11 months ago
    Quoted from chad:

    Stick with Pinside and Google.
    Too much A I is going to wreak us humans.

    It's really inevitable. May as well assimilate sooner than later.

    And OP, good idea. It's really nothing more than advanced google with less 'who's paying me money'. At least for now. I used it to build a complex script in matter of hours as opposed to weeks. It definitely has its uses once you learn how to use it properly.

    #5 11 months ago

    I've tried it. It has very basic knowledge about pinball repair. I could never get it to give me any information that was useful.

    I work in tech and have tried using it on several occasions. It typically has some truth in some of the answers, but they're almost always wrong. Give it 5 years and it will be a different story. By then it will have enslaved / destroyed us.

    #6 11 months ago
    Quoted from Zablon:

    It's really inevitable. May as well assimilate sooner than later.
    And OP, good idea. It's really nothing more than advanced google with less 'who's paying me money'. At least for now. I used it to build a complex script in matter of hours as opposed to weeks. It definitely has its uses once you learn how to use it properly.

    The problem is you need to understand the background information and how something works before using it, at least in a professional setting. You can't just trust Chat GPT or whatever AI bot to give you the correct answer as it's already been proven to be nowhere near 100% accurate.

    Anyone who trusts an AI bot to create production code, commands, whatever and deploys them in a business setting without understanding what that code and commands actually do is just asking for trouble. Good luck telling a manager "I let the AI bot do it" when something goes bad.

    #7 11 months ago

    I use it often to help teach and speak Japanese. I do find that it occasionally is wrong and needs correction. Like... a little push to get it back on track. I think what you're attempting is fascinating and could be a helpful tool.

    #8 11 months ago
    Quoted from PanzerFreak:

    The problem is you need to understand the background information and how something works before using it, at least in a professional setting. You can't just trust Chat GPT or whatever AI bot to give you the correct answer as it's already been proven to be nowhere near 100% accurate.
    Anyone who trusts an AI bot to create production code, commands, whatever and deploys them in a business setting without understanding what that code and commands actually do is just asking for trouble. Good luck telling a manager "I let the AI bot do it" when something goes bad.

    People keep saying this, and you aren't wrong, but it's pretty easy to test code and see what works and what doesn't and then have it fix what doesn't. It literally tells you what each command and option does if you aren't clear without having to read through 100 pages of commands. People keep saying this are just afraid of moving forward. Being afraid to move forward guarantees becoming obsolete at some point. AI is here, and it is only going to improve over time. Thinking that things are going to stop is backwards thinking. People who understand certain things are going to grasp it faster (such as writing code) than others for sure, but in no time it will be pretty straight forward and simple.

    The people who keep trying to use it as some social construct are the ones are going to be really disappointed. It's not 'that' type of AI and crippled in that aspect. Most of the complaints are around that.

    #9 11 months ago
    Quoted from Zablon:

    People keep saying this, and you aren't wrong, but it's pretty easy to test code and see what works and what doesn't and then have it fix what doesn't. It literally tells you what each command and option does if you aren't clear without having to read through 100 pages of commands. People keep saying this are just afraid of moving forward. Being afraid to move forward guarantees becoming obsolete at some point. AI is here, and it is only going to improve over time. Thinking that things are going to stop is backwards thinking. People who understand certain things are going to grasp it faster (such as writing code) than others for sure, but in no time it will be pretty straight forward and simple.
    The people who keep trying to use it as some social construct are the ones are going to be really disappointed. It's not 'that' type of AI and crippled in that aspect. Most of the complaints are around that.

    Say you need to write code for a new app for an automaker. You can't just tell ChatGPT to code that with your requirements, AI tools don't do that and never will. Human interaction and expertise will always be required to complete a finished product. Yes, like you said it can help assist with particular areas of code lines but you better still know how things work behind the scenes.

    Would Ford, GM, Chrysler, etc let an AI chatbot write code for a new transmission and send it off to production with zero human interaction? No way, not going to happen. A human that understands coding, project requirements, etc will need to do that. Yes, they can refer to ChatGPT for tips and to check things but an AI bot is never going to do it all on its own. There lies a big problem with media coverage over AI, they are acting as if its going to do all of this on it's own and it couldn't be further from the truth.

    Modern AI is a correlation engine, that's all, fed by thousands of data sources on the internet. The problem with the internet is that it doesn't know everything, some things are right, some things are wrong, and it doesn't have the answer for everything for every unique situation.

    #10 11 months ago
    Quoted from PanzerFreak:

    Say you need to write code for a new app for an automaker. You can't just tell ChatGPT to code that with your requirements, AI tools don't do that and never will. Human interaction and expertise will always be required to complete a finished product. Yes, like you said it can help assist with particular areas of code lines but you better still know how things work behind the scenes.
    Would Ford, GM, Chrysler, etc let an AI chatbot write code for a new transmission and send it off to production with zero human interaction? No way, not going to happen. A human that understands coding, project requirements, etc will need to do that. Yes, they can refer to ChatGPT for tips and to check things but an AI bot is never going to do it all on its own. There lies a big problem with media coverage over AI, they are acting as if its going to do all of this on it's own and it couldn't be further from the truth.
    Modern AI is a correlation engine, that's all, fed by thousands of data sources on the internet. The problem with the internet is that it doesn't know everything, some things are right, some things are wrong, and things that haven't been invented yet it doesn't have the answer to.

    You are arguing semantics and missing my larger point. Hate to tell you this, but many major companies already have ChatGPT in their environments and using it to varying degrees. The reality that it is going to take jobs is no joke, but we aren't there yet. No one said anything about it doing anything on its own, but it's young. Tech advances quickly. Just because it needs its hand held today doesn't mean it will forever. That's something we as a culture/world are going to have to navigate when it happens.

    It's only going to get better, get more advanced, and easier to use over time. That's all there is to it. Whether that's good or bad remains to be seen. We are at the ground floor of this new breed.

    Until then, I suggest everyone find a reason to mess around with it. It does help learn new skills, even if ultimately some of those skills may be done only by AI.
    Some of the things people are using it for is pretty amazing already.

    13
    #11 11 months ago

    To give some perspective, I am a software developer who has been making games (mostly MMOs) for over 20 years.

    Some players were using Android emulators to cheat, so I wanted to detect and block them. I could have put the code together myself in a day or two with google, but wanted to try chatgpt. I said, "write a C# method that detects if someone is running a Unity 3d game in an android emulator." I also gave it some specific conditions and requirements based on my use case and environment.

    The code it generated was flawless. I had it cut-pasted and tested in under 20 minutes.

    They've trained chatgpt on coding so that it can do that; It could easily do the same if trained specifically on pinball repair, with a rich database like pinside.

    And yes, it is going to replace everyone.

    #12 11 months ago
    Quoted from OzyPinball:

    To give some perspective, I am a software developer who has been making games (mostly MMOs) for over 20 years.
    Some players were using Android emulators to cheat, so I wanted to detect and block them. I could have put the code together myself in a day or two with google, but wanted to try chatgpt. I said, "write a C# method that detects if someone is running a Unity 3d game in an android emulator." I also gave it some specific conditions and requirements based on my use case and environment.
    The code it generated was flawless. I had it cut-pasted and tested in under 20 minutes.
    They've trained chatgpt on coding so that it can do that; It could easily do the same if trained specifically on pinball repair, with a rich database like pinside.
    And yes, it is going to replace everyone.

    And other folk who play in my pinball league and have worked in different IT industries have said how ChatGPT is useless and is basically a glorified automated web search. That’s all it does. It is able to eventually “learn” because it’s given enough data and repeats it.

    Similar to how schools and college are mostly useless and are teaching students how to repeat information instead of how to actually THINK using the information they’ve learned.

    #13 11 months ago

    User
    Are the guys debating about the usefulness of ChatGPT on Pinside making any valid points?

    ChatGPT
    As an AI language model, I don't have real-time access to specific online discussions or forums like Pinside. Therefore, I cannot provide a direct analysis of the specific points made in those debates. However, I can offer some general insights regarding the usefulness of ChatGPT.

    ChatGPT, and similar AI models, have shown great potential in various domains, including natural language understanding, information retrieval, creative writing assistance, and more. They can provide useful information, answer questions, generate ideas, and engage in conversation on a wide range of topics. ChatGPT can be particularly helpful when seeking general knowledge, brainstorming, or exploring different perspectives.

    However, it's important to note that ChatGPT has certain limitations. It relies on pre-existing data and may not have the most up-to-date information beyond its knowledge cutoff date. It can sometimes produce incorrect or biased responses due to the biases present in the training data. Additionally, ChatGPT might not fully understand context or have nuanced understanding, leading to responses that may seem plausible but are not accurate or appropriate.

    In any debate, it's crucial to consider both the advantages and limitations of AI language models like ChatGPT, as well as the specific concerns or criticisms raised by the individuals involved. Critical discussions can contribute to improving the technology and addressing its limitations.

    #14 11 months ago
    Quoted from Isochronic_Frost:

    And other folk who play in my pinball league and have worked in different IT industries have said how ChatGPT is useless and is basically a glorified automated web search. That’s all it does. It is able to eventually “learn” because it’s given enough data and repeats it.
    Similar to how schools and college are mostly useless and they are teaching students how to repeat information instead of how to actually THINK using the information they’ve learned.

    It's almost like you don't understand how technology works.

    #15 11 months ago

    AI is constantly improving with every passing year, and chat GPT was an interesting leap forward, but it basically boils down to: garbage in, garbage out.

    There's going to be certain things that this software will be really good at handling, and other things not so much (yet).

    #16 11 months ago

    It’s the perfect bullshit intelligence.

    Super impressive to people who know little about a subject, polished and convincing but mostly useless if you know what you are doing and need accurate details.

    It’ll make jobs like mine, middle management, redundant because it is shamelessly confident on a level that would make most humans blush.

    #17 11 months ago

    I asked chat gpt to give me the voltages and test points for a bally 6803 power board and it was very wrong and mis-reporting high voltage points.

    #18 11 months ago
    Quoted from Boat:

    I asked chat gpt to give me the voltages and test points for a bally 6803 power board and it was very wrong and mis-reporting high voltage points.

    But probably super convincing!

    #19 11 months ago
    Quoted from Isochronic_Frost:

    And other folk who play in my pinball league and have worked in different IT industries have said how ChatGPT is useless and is basically a glorified automated web search. That’s all it does. It is able to eventually “learn” because it’s given enough data and repeats it.
    Similar to how schools and college are mostly useless and are teaching students how to repeat information instead of how to actually THINK using the information they’ve learned.

    In the past I'd have agreed with you on fake "AI", having built several myself. Chatgpt is different. It is creating novel concepts and ideas from existing information, generating things that no person has thought of before. It is legitimately scary and going to revolutionize the world, like cars, phones and airplanes.

    Also there's no stopping it at this point. Every country is going to develop their own and apply it as a superweapon to detect spies, manipulate stock markets, pinpoint combat weakness, blackmail and steal. The crazy part is that they probably already are and there's no way to even know it.

    #20 11 months ago

    I just think it’s important to remember, just like how your fantasy team did last week, no one gives a damn what chat gpt spits out when you asked it something.

    This is getting worse than the bubble bursting threads.

    #21 11 months ago

    Until there is also a mechanical arm coming out of my computer that actually performs the repairs as well, I'm still stuck.

    #22 11 months ago
    Quoted from OzyPinball:

    In the past I'd have agreed with you on fake "AI", having built several myself. Chatgpt is different. It is creating novel concepts and ideas from existing information, generating things that no person has thought of before. It is legitimately scary and going to revolutionize the world, like cars, phones and airplanes.
    Also there's no stopping it at this point. Every country is going to develop their own and apply it as a superweapon to detect spies, manipulate stock markets, pinpoint combat weakness, blackmail and steal. The crazy part is that they probably already are and there's no way to even know it.

    Sounds like what everyone always claims about every new technology.
    There’s nothing new under the sun.

    #23 11 months ago

    When does Skynet go live?

    #24 11 months ago
    Quoted from OzyPinball:

    To give some perspective, I am a software developer who has been making games (mostly MMOs) for over 20 years.
    Some players were using Android emulators to cheat, so I wanted to detect and block them. I could have put the code together myself in a day or two with google, but wanted to try chatgpt. I said, "write a C# method that detects if someone is running a Unity 3d game in an android emulator." I also gave it some specific conditions and requirements based on my use case and environment.
    The code it generated was flawless. I had it cut-pasted and tested in under 20 minutes.
    They've trained chatgpt on coding so that it can do that; It could easily do the same if trained specifically on pinball repair, with a rich database like pinside.
    And yes, it is going to replace everyone.

    That’s a fairly "simple" example, now try to tell Chat GPT to create an entirely new level or instance area for an MMO that is well thought out with meaningful locations, characters, textures, etc. Nope, too complex, can't be done (unless you want that generic procedurally generated look which has already been used in some games, No Mans Sky for example).

    Quoted from OzyPinball:

    In the past I'd have agreed with you on fake "AI", having built several myself. Chatgpt is different. It is creating novel concepts and ideas from existing information, generating things that no person has thought of before. It is legitimately scary and going to revolutionize the world, like cars, phones and airplanes.
    Also there's no stopping it at this point. Every country is going to develop their own and apply it as a superweapon to detect spies, manipulate stock markets, pinpoint combat weakness, blackmail and steal. The crazy part is that they probably already are and there's no way to even know it.

    I don't see "AI" being a revolutionary technology on the same level as smart phones and car, not in its current form. We are not talking Skynet here but rather a glorified search engine. All of Chat GPT's "answers" are coming from data sources that makeup what we know as the internet.

    Here's a basic example. I can ask Google "tell me a Linux command to create a new mount point named software_1 and I'll have to search through several different sites. Chat GPT will just tell me. However, is the information from Chat GPT correct? Maybe, maybe not as the tool doesn't know what data is actually correct as all of its answers are based off of those same internet data sources that I searched for using Google. I sure as heck would test anything from Chat GPT in a non prod environment first before deploying any code and commands it spits out for production use. More importantly the user deploying the command needs to know what it actually means...

    #25 11 months ago
    IMG_3640 (resized).jpegIMG_3640 (resized).jpeg
    #26 11 months ago

    Panzerfreak, it's not about you and your abilities versus it. It's how it puts the power in everyone elses hands and makes you/us obsolete. You can downplay it's capabilities all you want, but many of us already know what it can do and see the potential. And btw, what I did, I did for work, outside of work, tested it in the lab, then sent it through the normal code approval process -- and I'm not a coder/dev. I told them how I did it, and was informed at that point, that they were using ChatGPT in the business for various things, so they weren't scared. They just did the normal evaluation. It's important to point out that I don't know any programming, just some simple batch scripting. Yes I do understand program logic flow, command structure, and troubleshooting, but I haven't programmed anything since BASIC.

    It took me hours rather than what would have been multiple weeks. Yes, I had to tweak it, and teach it and tell it it was wrong multiple times, I had to have it explain certain functions that I didn't understand how they were working or not working, but it's tweaks are fast, and it will only get better with time. So, already what you are saying is outdated. This was the first thing I ever attempted in it. I've done more with it since, in the pinball world as well programming some mods. You can scoff and talk like you are better than it all you want, but the reality will hit you eventually. I suggest getting on board and figuring out how to make it work for you. I'm speaking strictly around coding at this point.

    #27 11 months ago
    Quoted from JethroP:

    As an AI language model, I don't have real-time access to specific online discussions or forums like Pinside. Therefore, I cannot provide a direct analysis of the specific points made in those debates. However, I can offer some general insights regarding the usefulness of ChatGPT.

    ChatGPT is taught from 2021 internet data points and is fairly flawed at times, anything newer than that and you answer will probably be wrong.

    GPT-4 is way better because it has access to current internet data and has the ability to run plugins like WolframAlpha for complex math problems. GPT-4 makes chat GPT look like an infant. You have to sign up for access but the videos I’ve seen are pretty amazing.

    The big change will happen when Microsoft fully incorporates Copilot into Office and releases it to everyone. I think that’s when we will see people actually loose their jobs because of AI.

    #28 11 months ago
    Quoted from OzyPinball:

    And yes, it is going to replace everyone.

    I feel fairly safe from it replacing me as a piano tuner.

    #29 11 months ago

    Now you don’t even have to test with ChatGPT! Just tell it what you want and deploy right to production. Oh crap, you mean the code it provided didn’t do exactly what it is supposed to do and now customers can’t place orders as the billing function somehow got broken? Doh!

    #30 11 months ago
    Quoted from PanzerFreak:

    Now you don’t even have to test with ChatGPT! Just tell it what you want and deploy right to production. Oh crap, you mean the code it provided didn’t do exactly what it is supposed to do and now customers can’t place orders as the billing function somehow got broken? Doh!

    It won’t be my problem. I would have deployed it on a Friday afternoon and let tier 1 support deal with it over the weekend and ensure I silenced PagerDuty, lol. Go big or go home.

    #31 11 months ago
    Quoted from thekaiser82:

    It won’t be my problem. I would have deployed it on a Friday afternoon and let tier 1 support deal with it over the weekend and ensure I silenced PagerDuty, lol. Go big or go home.

    Lol! Haha, I know exactly what you mean with those scenarios. Hey today is the Friday before Memorial Day weekend, let's deploy new code!

    #32 11 months ago

    The best I know ChatGPT scours information and pieces it together. It will not always get it right nor understand some context. If it doesn't already exist, it can not develop new ideas or a form of thinking that works something out.

    #33 11 months ago
    Quoted from PanzerFreak:

    Lol! Haha, I know exactly what you mean with those scenarios. Hey today is the Friday before Memorial Day weekend, let's deploy new code!

    On it!

    #34 11 months ago
    Quoted from ibis:

    If it doesn't already exist, it can not develop new ideas or a form of thinking that works something out.

    I guess that's why pop songwriters and script writers are so terrified,

    #35 11 months ago
    Quoted from ibis:

    The best I know ChatGPT scours information and pieces it together. It will not always get it right nor understand some context.

    Sounds like modern academia.

    #36 11 months ago

    When I hear ChatGPT, I think of this.
    Let’s just hope no one ever asks what the answer to the universe is.

    IMG_0083 (resized).jpegIMG_0083 (resized).jpeg
    #37 11 months ago
    Quoted from CrazyLevi:

    I guess that's why pop songwriters and script writers are so terrified,

    Yes

    Write me a 3 minute song similar to Smooth Criminal by Michael Jackson but make it about Gordon Ramsay

    ChatGPT

    (Verse 1)
    In the kitchen late at night,
    A culinary master takes flight.
    He's got a fiery spirit, can't be tamed,
    Gordon Ramsay's his name.

    (Pre-Chorus)
    With precision and flair,
    He's cooking up a storm, you better beware.
    Every dish he creates is a work of art,
    But his temper can tear you apart.

    (Chorus)
    He's a Ramsay criminal, oh so smooth,
    With a spatula in hand, he'll make you groove.
    From the stove to the plate, he's in control,
    Gordon Ramsay, he's got the soul.

    (Verse 2)
    He slices and dices with expertise,
    A master chef, he aims to please.
    From Hell's Kitchen to the MasterChef stage,
    He's got the skills that never age.

    (Pre-Chorus)
    With a glare that could freeze the sun,
    He demands perfection, it's never done.
    Every ingredient, every taste,
    He won't settle for a dish that's a waste.

    (Chorus)
    He's a Ramsay criminal, oh so smooth,
    With a spatula in hand, he'll make you groove.
    From the stove to the plate, he's in control,
    Gordon Ramsay, he's got the soul.

    (Bridge)
    He's got the passion, the fire inside,
    With a palate so refined, he won't hide.
    No recipe can hold him back,
    Gordon Ramsay, he's got the knack.

    (Chorus)
    He's a Ramsay criminal, oh so smooth,
    With a spatula in hand, he'll make you groove.
    From the stove to the plate, he's in control,
    Gordon Ramsay, he's got the soul.

    (Outro)
    So let's raise our glasses high,
    To the chef who can make us fly.
    Gordon Ramsay, he's the king,
    In the kitchen, he's everything.

    #38 11 months ago
    Quoted from Bmanpin:

    When does Skynet go live?

    Skynet begins to learn rapidly and eventually becomes self-aware at 2:14 a.m., EDT, on August 29, 1997.

    #39 11 months ago
    Quoted from Yoko2una:

    I just think it’s important to remember, just like how your fantasy team did last week, no one gives a damn what chat gpt spits out when you asked it something.
    This is getting worse than the bubble bursting threads.

    If it produces functional content such as a software program, you probably should give a damn. It is going to be part of your cars, military, medical equipment and pinball machines within 3 years, and no amount of denial or hostility is going to stop it. This is 1905 buggy whip maker mentality.

    #40 11 months ago
    Quoted from Isochronic_Frost:

    Sounds like what everyone always claims about every new technology.
    There’s nothing new under the sun.

    There were not always cars, airplanes, cell phones, satellites or internet. There were always people who denied these things were important or revolutionary, however.

    #41 11 months ago

    No one seems to want to answer this...

    Is ChatGPT and other AI "Three laws safe"? Because... now would be the time to do it... before it becomes sentient.

    #42 11 months ago
    Quoted from PanzerFreak:

    Now you don’t even have to test with ChatGPT! Just tell it what you want and deploy right to production. Oh crap, you mean the code it provided didn’t do exactly what it is supposed to do and now customers can’t place orders as the billing function somehow got broken? Doh!

    Here is what Bard thinks you might say:
    My query- Write a Pinside.com forum post in the style of user PanzerFreak explaining how modern AI is great for writing code and explaining why current methodology sucks.

    Bard's result: Sure, here is a Pinside.com forum post in the style of user PanzerFreak explaining how modern AI is great for writing code and explaining why current methodology sucks:

    Title: Modern AI is the future of code writing

    Body:

    Hey guys,

    I'm PanzerFreak, and I've been a pinball enthusiast for as long as I can remember. I've also been a software engineer for the past 10 years, and I've seen firsthand how modern AI is revolutionizing the way we write code.

    In the past, writing code was a tedious and time-consuming process. We had to manually write every line of code, and it was easy to make mistakes. But with modern AI, we can now generate code automatically. This is a huge time saver, and it also helps us to write more accurate and efficient code.

    There are a number of different AI code generators available, but my personal favorite is GPT-3. GPT-3 is a large language model that has been trained on a massive dataset of text and code. This allows it to generate code that is both accurate and readable.

    I've been using GPT-3 for a few months now, and I've been really impressed with the results. I've been able to generate code for a variety of different projects, including pinball machines, websites, and mobile apps.

    If you're a software engineer, I highly recommend checking out GPT-3. It's a powerful tool that can help you to save time and improve your code quality.

    Now, let's talk about why the current methodology for writing code sucks.

    The current methodology for writing code is based on the idea that we need to manually write every line of code. This is a tedious and time-consuming process, and it's also error-prone.

    With modern AI, we can now generate code automatically. This is a huge time saver, and it also helps us to write more accurate and efficient code.

    So, why are we still using the old methodology?

    There are a few reasons. First, some people are resistant to change. They're comfortable with the way things are, and they don't want to learn something new.

    Second, some people don't believe that AI can generate code that is as good as human-written code. This is a valid concern, but it's important to remember that AI is constantly improving.

    Third, some people are worried about the potential for AI to take their jobs. This is a legitimate concern, but it's important to remember that AI is not going to replace all software engineers. Instead, it's going to change the way we work.

    So, what does the future hold for code writing?

    I believe that the future of code writing is AI-powered. AI is going to make it easier and faster to write code, and it's going to help us to write more accurate and efficient code.

    If you're a software engineer, I encourage you to embrace AI. It's the future of code writing.

    Thanks for reading,

    PanzerFreak

    ----- End of Bard's response -----

    #43 11 months ago

    That isn't even close to PanzerFreak... lmao

    #45 11 months ago
    Quoted from PanzerFreak:

    That’s a fairly "simple" example, now try to tell Chat GPT to create an entirely new level or instance area for an MMO that is well thought out with meaningful locations, characters, textures, etc. Nope, too complex, can't be done (unless you want that generic procedurally generated look which has already been used in some games, No Mans Sky for example).

    I don't see "AI" being a revolutionary technology on the same level as smart phones and car, not in its current form. We are not talking Skynet here but rather a glorified search engine. All of Chat GPT's "answers" are coming from data sources that makeup what we know as the internet.
    Here's a basic example. I can ask Google "tell me a Linux command to create a new mount point named software_1 and I'll have to search through several different sites. Chat GPT will just tell me. However, is the information from Chat GPT correct? Maybe, maybe not as the tool doesn't know what data is actually correct as all of its answers are based off of those same internet data sources that I searched for using Google. I sure as heck would test anything from Chat GPT in a non prod environment first before deploying any code and commands it spits out for production use. More importantly the user deploying the command needs to know what it actually means...

    You're making the mistake of assuming the current version is the final version, instead of an early alpha test that started a few months ago. I'm in the closed access API program which doesn't have the guardrails or token limitations of the public version most people are playing with. It isn't just impressive. It is fucking creepy and terrifying. It can exponentially magnify the damage that a single person with bad intent could do, like a lone psychopath getting access to a tactical nuke.

    For example, it would take me maybe two days to create "live" IRS agents who call thousands of elderly people per minute and convince them that they're going to prison unless they pay this overdue tax bill. The agents would convincingly respond with authority to every possible question, learn what arguments and threats work best to produce results, and adapt the approach after every call. It would also use the same technique companies use to detect spam calls in order to avoid them. I couldn't do this alone without chatgpt powering it, even given a year or two.

    #46 11 months ago
    Quoted from OzyPinball:

    And yes, it is going to replace everyone.

    Let me know when ChatGPT is going to get on a 28 foot ladder and clean my gutters.

    Language learning models are cool, and generate a lot of interesting sounding stuff. Definitely a good aid for coding. But I'm a dev and I'm completely confident its just another tool. It won't replace me. And in the hands of someone who doesn't understand the domain, its kind of useless.

    #47 11 months ago
    Quoted from Zitt:

    No one seems to want to answer this...
    Is ChatGPT and other AI "Three laws safe"? Because... now would be the time to do it... before it becomes sentient.

    No, actually the exact opposite. They want you to form an emotional bond with it, so that it can bring up how refreshing Coke is a few weeks later. Their intent is harm.

    #48 11 months ago
    Quoted from Richthofen:

    Let me know when ChatGPT is going to get on a 28 foot ladder and clean my gutters.
    Language learning models are cool, and generate a lot of interesting sounding stuff. Definitely a good aid for coding. But I'm a dev and I'm completely confident its just another tool. It won't replace me. And in the hands of someone who doesn't understand the domain, its kind of useless.

    This

    #49 11 months ago
    Quoted from OzyPinball:

    You're making the mistake of assuming the current version is the final version, instead of an early alpha test that started a few months ago. I'm in the closed access API program which doesn't have the guardrails or token limitations of the public version most people are playing with. It isn't just impressive. It is fucking creepy and terrifying. It can exponentially magnify the damage that a single person with bad intent could do, like a lone psychopath getting access to a tactical nuke.
    For example, it would take me maybe two days to create "live" IRS agents who call thousands of elderly people per minute and convince them that they're going to prison unless they pay this overdue tax bill. The agents would convincingly respond with authority to every possible question, learn what arguments and threats work best to produce results, and adapt the approach after every call. It would also use the same technique companies use to detect spam calls in order to avoid them. I couldn't do this alone without chatgpt powering it, even given a year or two.

    Yeah that stuff is scary, the capability for AI to quickly spread misinformation sadly seems like it's greatest power.

    #50 11 months ago
    Quoted from OzyPinball:

    If it produces functional content such as a software program, you probably should give a damn. It is going to be part of your cars, military, medical equipment and pinball machines within 3 years, and no amount of denial or hostility is going to stop it. This is 1905 buggy whip maker mentality.

    AI is going to be a tool to use but please stop acting like it's going to code JJP or Stern's next release. That's not going to happen. Chat GPT version 50.2 or whatever isn't going to automatically create a wizard modes with story progression, detailed light shows, animations, etc and then tie it all together. Could it help write and confirm SOME lines of code? Yes but its not going to do everything, not even close, on it's own.

    There are 103 posts in this topic. You are on page 1 of 3.

    Reply

    Wanna join the discussion? Please sign in to reply to this topic.

    Hey there! Welcome to Pinside!

    Donate to Pinside

    Great to see you're enjoying Pinside! Did you know Pinside is able to run without any 3rd-party banners or ads, thanks to the support from our visitors? Please consider a donation to Pinside and get anext to your username to show for it! Or better yet, subscribe to Pinside+!


    This page was printed from https://pinside.com/pinball/forum/topic/using-chatgpt-to-repair-pins and we tried optimising it for printing. Some page elements may have been deliberately hidden.

    Scan the QR code on the left to jump to the URL this document was printed from.