Sharing my ChatGPT Bot trained to help with Gen4 v8 e38 engines.

[crunchybeans]

Hey Guys,

I've been mostly on the asking side of things here, soaking up the wisdom this forum generously offers. Today, I'm excited to flip the script a bit and share something from my end that I hope can contribute back to our community.

I've developed a GPT bot specifically tailored for tuning Gen 4 V8 E38 engines. This isn't your average chatbot. It's been trained with a focus on the intricacies of our beloved engines, delving into the nuances of tuning that we all wrestle with. Whether it's VE tables, spark advance, or the ever-pesky fuel trims, this bot's got insights.

Link to GPT Bot: https://bit.ly/49ojd5M

What sets it apart is its ability to process and analyze tuning data, providing recommendations and adjustments that align with our tuning goals. It's like having a tuning buddy by your side, one that doesn't drink your beer or borrow your tools.

I've been testing it out on my own projects, tweaking it based on real-world results, and I'm finally at a point where I feel it can genuinely offer value to others in this space. It's not just about throwing numbers and tables at you; it's about understanding the why behind the tuning decisions we make. I have been working with it lately in optimizing the math for GMVE tuning VVE, as well as optimizing my filtering formulas. It is trained with all of the tables and functions for my engine, and able to answer the question: "What table do I need to edit or look at if I want to ..."

Here's a glimpse of what it can do:

Analyze VE tables and suggest optimizations
Offer advice on managing air/fuel ratios across different loads and RPMs
Provide insights on maximizing efficiency and performance through timing adjustments

I'm keen to get your thoughts, feedback, and, honestly, any tuning puzzles you might want to throw its way. I believe in the power of this community to push the envelope on what's possible with our engines, and I'm hoping this bot can be a tool in our collective tuning arsenal.

Looking forward to hearing what you all think and seeing how we can evolve our tuning strategies together.

-Crunchybeans

[blindsquirrel]

Dumb over-hyped shit. It takes more time to fact check the stupid chatbot for the inevitable errors than it would to just do the research yourself. The current AI thing is a scam, to collect venture capital funding for smooth-brained startups. Don't fall for it.

[hjtrbo]

Na fuck that. I'm not paying for it just to have a look.

And you've used this forum data to train it. I wouldn't trust 3/4 of the shit on here.

[sgod1100]

Dumb over-hyped shit. It takes more time to fact check the stupid chatbot for the inevitable errors than it would to just do the research yourself. The current AI thing is a scam, to collect venture capital funding for smooth-brained startups. Don't fall for it.

"Smooth brained"....lol. I love using that term...

[C5Diaz]

Seems like a cool concept, new technology will always get pushback from the purest out there. i'll check it out and give any feedback I can.

[hjtrbo]

The problem with it is, once it's scraped all the data off websites like this and been trained, then all the sheep follow, where does it go for new concepts and new knowledge? Having used chat gpt for some coding help, I can tell you the model can easily get itself twisted up and spit out absolute junk.

[Tunercharged]

I think that these ?AI to replace learning concepts? are not beneficial. Look up the question on the internet and while shifting through answers maybe figure out what you really need to learn. If we just trust this corporate stuff then ask ourselves why can we improve on their initial designs and processes? We need to not be boxing ourselves into mainstream answers which is all that AI will provide. The future is now! Wait? now!

[TheMechanic]

Been doing the "computer" thing for over 40 years back to DOS.
Garbage in Garbage out.
Unless you have pure 100% infallible, verified data you will have mistakes. Here is were experience will over rule a computer output. Also everything is seat of the pants subjective. Everyone will have a different "opinion" of an improvement. Sometimes that improvement is in the head of the person who WANTS the outcome to be good instead of it actually being an improvement.
Only way I can see AI creating improvement is to have an accurate engine dyno that with every possible data sensor running through every billion possible combinations of changes and the improvement it has. Next is power improvement compared to longevity. Drag racing see's leaning out the mixture at the top end that improves power and damages the engine. Maybe someday but sure as hell not today.

[blindsquirrel]

The Navy revealed the embryo of an electronic computer today that it expects will be able to walk, talk, see, write, reproduce itself and be conscious of its existence.

New York Times, July 8, 1958

But maybe this time it's for real?

[PBJ83CAMARO]

Someone got no wrinkles........

[blindsquirrel]

Machine-generated text from an Ebay listing:

This 1963 Chevrolet Nova is a true original and has been under the same ownership for more than 5 years. The vehicle is fully drivable and has been well-maintained throughout its lifetime. The exterior is a beautiful brown color and the title has been rebuilt for added peace of mind. This classic car is perfect for any collector or enthusiast looking to add a piece of automotive history to their collection.

It can't spell 'driveable', it also doesn't know what a 'title' is, or that a 'rebuilt title' is NOT a good thing the same way a 'rebuilt engine' or 'rebuilt suspension' would be. It doesn't know anything. It just strings words together so that they look plausible from a distance.

From a page on Gearstar's website:

A notch more powerful than the Transmission Level 2, Level 3 boasts 600 Horse Power that can generate up to 600 ft. lbs. of torque. This automatic transmission for sale is compatible with 10? custom built lock-up billet racing converters. As an automatic transmission gear, this has a stall level of 2600 to 3600 rpm.

WT actual F? That doesn't mean anything! All of that is just wrong on every conceivable level.

This Edelbrock Performer RPM Hydraulic Roller Camshaft is designed for use with Chevy SBC engines. It has a lift size of .539" and is compatible with interchange part numbers EP4001, 399M, and 700-121. The camshaft is made from high-quality materials and is built to last, ensuring that you can rely on it for years to come. It is a great choice for anyone looking to upgrade their engine's performance.

'It has a lift size of .539', said no living human who knows camshafts, ever. So it is made from high-quality materials, does that include SADI? Or is it 5140, or 8620? I do not think you know what you are talking about, I am not going to just take your word for it, Mr Brain-in-a-box (allegedly).

Made during the golden age of automobiles, this Impala is still running strong, and with only 1000 miles on the odometer, it's ready for you to make it your own. While it has been in the hands of its current owner for over 5 years, the vintage sedan is in need of some work before it can be driven regularly. With its timeless styling and iconic design, this Impala is perfect for the classic car enthusiast.

Running strong, but in need of some work before it can be driven? (the car in this listing did not have an engine or transmission)

With this fuel injector adapter, you can easily customize your fuel injection system without worrying about the condition of the item. It is designed to provide a seamless upgrade for your Chevy LS Truck, delivering improved performance and efficiency. Whether you're a professional mechanic or a DIY enthusiast, this adapter is sure to meet your needs.

This is The Future? This is what has inspired thousands of breathless news headlines over the past year? This dumb shit?? This is what it feels like to go insane, isn't it?

[gtstorey]

I always assumed this stuff was written by people that can't actually compose sentences without emojis or abbreviations along with having no actual technical knowledge at all.

And then there are the translations from other languages. Instructions that are translated from Chinese into English have sounded like AI for as long as I can remember.

[Nebulazer03]

Hey Crunchy Beans,

Dont let these guys discourage your idea. I get where youre coming from, and I think your ideas might just be landing in front of the wrong crowd here. I think its a great idea. In fact, Ive built the same chatbot for myself (I also have a V8 E38 LS3). A lot of people, myself included, have used similar AI tools, and its proven to save time and minimize errors. Its not advanced enough to do all the work for you, but it definitely takes some of the load off.

For anyone else reading this who doesnt agree, Id like to point out that past attempts with computer programs built by someone else to tune were inefficient or failed because computer programs from back in the day are never going to be more powerful than human brains.

But heres the thing? Artificial intelligence is not a computer program that someone has predetermined how it will work; it is not pre-programmed. AI has immediate access to the knowledge of the entire internet at its fingertips. Its been trained to connect relevant information from all over the web and make sense of it, almost like building a spider web of knowledge in real time. So, when someone here says, Its quicker to research it yourself than to wait for an answer, theyre wrong. The model already knows all the relevant threads, forums, and posts within seconds?stuff wed spend hours digging through manually. Its faster and more efficient than we can be. The only advantage we have that it never will is the ability to learn in person or offline (e.g., training, experience, mentorship, private books, private courses). This model can capture images and text, comprehend the information, and translate what it means faster than we ever could.

And just to give you all a little more peace of mind, Im about to finish my bachelors degree in Artificial Intelligence, so trust me when I say Ive spent plenty of time with these tools and know what Im talking about. The only issue in play right now is time, or the lack of it. ChatGPT has a seemingly infinite amount of knowledge, which means it started out with a near-infinite amount of inaccuracies. Its our job to keep asking it questions, pointing out its errors, and it will pick up the rest along the way. Like any tool, the more its refined and used, the better it gets over time. This is a new concept, and it needs time to become nearly perfect. As of today, it has already perfected many things people want it to do (advanced math, image comprehension, even human speech?search OpenAI advanced voice).

What you might not realize is that even some of the biggest companies out there, or big private companies that businesses work with, have been using artificial intelligence/machine learning to maintain, improve, and create their systems. These systems handle huge amounts of data and need machines to make accurate predictions. ChatGPT gives everyday people the option to create their own AI without needing to know much of anything. They lead the industry with the most researched and developed models. Its all public and open-source, super user-friendly, free, and its not for profit, so theres no hidden corporate agenda pushing this on people. It has already changed so much of the industry.

Every need will eventually be filled. If tuning is an option, its only a matter of time before youll hear of a product or service that can do it. And some lucky person is going to capitalize on that because they can meet the demand. The more supply they provide, the higher the demand will be.

Now, respectfully, dont get stuck in the past. Times have changed and will continue to do so. Just because the news headlines talk about AI doesnt mean its all hype pushed by big companies for profit?hating something because of the news is still being influenced by it. Just as you thought cell phones and computers were great ideas back in the day, weve now reached the end of that era of advancement (Google Moores Law). Now weve moved on to something new and exciting, so be excited!

Cheers!

[blindsquirrel]

It takes longer to fact-check the hallucination-prone chatbot than it would to just do the research yourself. Unless you want to save time by not fact-checking, in which case you get exactly what you deserve.

[gametech]

It takes longer to fact-check the hallucination-prone chatbot than it would to just do the research yourself. Unless you want to save time by not fact-checking, in which case you get exactly what you deserve.

A few lawyers have been fired and maybe disbarred already for not checking the results of this crap, so at least some good is coming from chatgpt, lol.

[abc]

"ChatGPT is the end result of someone assuming that if they gather up a large enough pile of sh*t it will spontaneously generate a pony."

Is the above the someone?

I'm excited to see where AI goes, sort of like watching the blower pop off of John Force's top fuel funny car, waiting with excitement to see where it lands and praying no one gets hurt.
When your dealing in absolutes, being able to sort fact from fiction has proven to be a problem for AI currently. Unfortunate for you and your indoctrination your gonna have a hard time convincing this crowd.

[blindsquirrel]

AI can do lots of stuff. It can't do much of anything reliably or well enough to be trusted. Will it eventually be able to, some day? Who knows. But right now it's a dumpster fire that's really good at convincing the gullible that it is magic. And also burning more cash than any tech invention ever in history. "AI" is a tool for generating the next round of investor funding.

[gametech]

The "my pillow" guy Mike Lindell's lawyers just recently delivered a brief in federal court that only had 30 defective citations in it, because of course they used AI to write it. I am assuming the streets would be safer with AI tuning due to all the non running engines lowering traffic volume.

[blindsquirrel]

This was written in mid-2023 but nothing of substance has changed since then. The marketing pitches have gotten slicker, is all.
https://softwarecrisis.dev/letters/llmentalist/

You give it text. It gives you a response that matches responses that texts like yours commonly get in its training data set.

Already, this is working along the same fundamental principle as the psychic's con: the LLM isn't "reading" your text any more than the psychic is reading your mind. They are giving you statistically plausible responses based on what you say. You're the one finding ways to validate those responses as being specific to you as the subject of the conversation.

Feynman: "The first principle is that you must not fool yourself, and you are the easiest person to fool."

side note: Have you had your credit/debit card declined recently? You know, stranded at the self-checkout in the grocery store with a cart full of stuff you're going to have to abandon right there and slink away in shame, because you have no way to pay for the stuff without a working card? Well you can thank "AI" for that. The dipshits at the card processors have fooled themselves into thinking that the "AI" they bought from the slick snake oil pitchman is smart enough to analyze transactions in real time and flag anything out of the ordinary. It can't do that, as evidenced by the cart full of groceries that are not going to be going home with you that day.

[pannetron]

blindsquirrel's reference is good. The *I* in AI is misleading! In 1999 and 2000, the cool *I* meant Internet and anxious investors piled in. Remember what happened? Today the *I* is different but anxious investors are piling in. The Large Language Models (LLM) scrape the publicly accessible Internet to load up their neural network. And since it's from the Internet, it must be true, right? DOGE wants access to government data because most of that is private so Musk's xAI can have data that no other LLM has.