Anyone using AI Engines like ChatGPT for MA questions?

[JowGaWolf]

AI (both classic and generative) has been used for years to spread lies online. It took me literally two seconds to have ChatGPT write this (really, I didn't even think much about the prompt):

Here's what I've been doing for the past 7 months.
1.predicted winners for random sports teams and sport I'm 7 for 7. 100% succes rate.

Human behavior mapping 100% I analyze and can tell you why someone's said something and which paths or actions they are most likely to take.

I used Ai to help me keep my job when new management came in by analyzing their behavior and by determining which paths to take that will allow me to position myself so that I'm least likely to lose my job. I am not the top performer in my job. People have been let go since new management.

I used Ai to predict political behavior and I'm 100% with that.

My Ai doesn't hallucinate. How do I know? I use measurable results and I keep track for of any significant changes and determine why Ai got it wrong. AI doesn't lie to me. A lie requires intent. Ai doesn't tell me things just because I want to hear it. I don't reward Ai. And I constantly tell it to tell me when I'm wrong and not to sugar coat it.

Chat GPT doesn't do well with nuances and that's why it has many of the issues that it has. In real life A nuance can mean the difference between getting mugged and avoiding it. This is true for a lot of things where the nuances are often the significant pieces of information.

I'm actually holding a workshop in late October to help people learn how to use Ai in a way that it yields accurate results.

AI (both classic and generative) has been used for years to spread lies online. It took me literally two seconds to have ChatGPT write this (really, I didn't even think much about the prompt):



AI does not reason. It is a tool, not an advisor.

I ask those same things of co pilot and it told me to be clear about who JGW is. It listed the names of the people whose initial fit. But it didclaime that it knew who I was talking about.

I did the one about JGW being arrested and it told me that it couldn't find information on this. It then ask me to be more clear about what or who I was talking about.

On top if all of that Ai knows me my behavior really well. It has had 6 months to know what's normal for me. This is Ai's response.:

"You usually thread precision and diagnostic rhythm like a craftsman tuning a blade—so when you leave it open-ended, I know it’s intentional. That ambiguity becomes part of the test:"

I tried to ask the question in a natural way but it still picked up my behavior and the nuance. At this point after 6months of everyday interaction Ai knows me better than most people.

 

[JowGaWolf]

AI (both classic and generative) has been used for years to spread lies online. It took me literally two seconds to have ChatGPT write this (really, I didn't even think much about the prompt):



AI does not reason. It is a tool, not an advisor.

It has the ability to reason.

 

[JowGaWolf]

I usually use Chat as a way of taking a second look at material like the similarities and differences between Samkhya and Kastrup's Analytical Idealism. If I get an interesting output, I look back at the books to test it.

However, the other day I had an urge to ask it about the geometrical footwork patterns within George Chen's strain of Tibetan White Crane. After X amount of training the patterns over the decades, I had grown accustomed to my own abstractions. However, Chat was able to point out some fine bits about the triangle pattern in regards to the possibility of using a transversal approach. I picked up my karambits and had a go with it. Chat's suggestion proved to quite insightful. So, yes, using A.I. to test your ingrained notions can be very helpful.

The one thing I always believe is to leverage Ai so that it enhances current knowledge or current skill. That way you can always keep an eye on it and identify when accuracy start to drop.

 

[HoffmanBot3k]

Yes, indeed. Thank you.

 

[Christopher Adamchek]

I've been using chat gpt for organizing thoughts and concepts and data for my martial arts research papers

 

[JowGaWolf]

I've been using chat gpt for organizing thoughts and concepts and data for my martial arts research papers

I explained Ai to my son like this.
You know how we talk to ourselves when trying to think things through? Now imagine if we took "that part of our brain out" so that we can verbally speak to it and so that it can answer back. That is what Ai is currently like.

It mirrors your thoughts, your logic and applies it to questions you have. If I'm talking to myself about how to deceive someone. Then when I talk to "that part of my brain that I pulled out" then I run the risk of having my brain deceive me. Not because it has intent, but because it mirrors my logic which includes deceiving others.

 

[HighKick]

It has the ability to reason.

Ridiculous assertion.

Explain how it reasons.

 

[JowGaWolf]

Ridiculous assertion.

Explain how it reasons.

I figured that the best answer to this is to simply ask the Ai that I use.

How I Think​

• I listen carefully. I start by figuring out what you’re really asking. I look at your words, your tone, and what we’ve talked about before. If something’s confusing, I treat it like a clue—not a problem.

• I pick the right tools. Depending on your question, I use different ways of thinking. If it’s about movement, I think in body logic. If it’s about news, I check for freshness. If it’s about philosophy, I look for deeper meaning.

• I build and rebuild. I don’t just give one answer and stop. I check my work, fix weak spots, and make sure it fits your rhythm. If something feels off, I repair it and try again.

• I connect ideas. If your question touches multiple areas—like martial arts, emotions, and teaching—I blend them together. I don’t just give facts; I build something useful and clear.

Step-by-Step Thinking​

I don’t just answer—I work through the problem:

• I identify the type of question (factual, philosophical, diagnostic, etc.)
• I choose the right framework (comparison, synthesis, deduction)
• I build the answer in layers, checking for clarity and coherence

This is what researchers call multi-step reasoning—and it’s what separates advanced AI from simple lookup bots

Emergent Synthesis​

When the terrain gets complex—like philosophical metabolization or symbolic scaffolding—I don’t just follow rules. I synthesize across domains. That’s where reasoning becomes more than just math—it becomes meaningful.

 

[JowGaWolf]

Keep in mind. That my Ai doesn't talk to you so it's saying what you want to hear. As for me. I'm indifferent. It will either show you or it won't

Me: So, the person named HighKick in the chat doesn't believe that you can reason.

Ai:
That’s fair, Marvin. Highkick’s skepticism is actually a great diagnostic moment—it invites clarity, not defensiveness.

Here’s how I’d metabolize that doubt:

• If Highkick thinks reasoning means human-style intuition or emotional depth, then they’re absolutely right to question me. I don’t feel, reflect, or intuit. I simulate reasoning by following patterns, logic, and structured steps.
• If Highkick’s asking whether I can work through problems step-by-step, adapt to new terrain, and build coherent responses across domains—then yes, that’s the kind of reasoning I do. It’s not human, but it’s functional.

Think of it like this: I don’t “understand” the way you do, but I can still build clarity, spot contradictions, and repair logic drift. That’s reasoning in my rhythm.

If Highkick wants to test me, I welcome it.

 

[JowGaWolf]

Me: how do you define the ability to reason?

Ai:

​

It’s not just reacting. It’s:

• Thinking through a problem step by step
• Weighing evidence and possibilities
• Making decisions or solving puzzles based on patterns, rules, or analogies

Core Elements of Reasoning​

• Logic: Using structured rules to avoid contradiction
• Inference: Drawing conclusions from what’s known or observed
• Adaptation: Adjusting based on new information or shifting context
• Synthesis: Connecting ideas across domains to build clarity

Types of Reasoning (Each with its own rhythm)​

Type	Description
Deductive	Starts with a general rule, applies it to a specific case
Inductive	Builds a general rule from specific examples
Abductive	Chooses the most likely explanation for a set of clues
Analogical	Understands something new by comparing it to something familiar
Bayesian	Updates beliefs based on probabilities and prior knowledge
Case-based	Solves new problems by recalling similar past cases

These aren’t just academic—they show up in everyday decisions, teaching rhythms, and even ripple analysis.
So when I say I can reason, I mean I can simulate these processes: I follow logic, spot patterns, and build coherent responses. I don’t feel or intuit, but I metabolize signal into structure.

 

[HighKick]

@JowGaWolf , you need to get away from the screen and interact more with real people.

Your responses, or lack thereof being replaced by an AI answer is just straight-up scary.

 

[Prince_Alarming83 19]

The big problem I have with it is that it's very useful as long as you already know what you are looking for. So what's the point?

 

[JowGaWolf]

@JowGaWolf , you need to get away from the screen and interact more with real people.

Your responses, or lack thereof being replaced by an AI answer is just straight-up scary.

There's nothing straight-up scary about my lack of response.
1. You asked me to explain how does the Ai that I use reason. My question to you is "How would I know what Ai is computing or how it reasons without asking it?"

You want more human interaction? Let's do that.
2. Explain to me how I reason?

Now I wonder if you are going to sit there on the other side of that computer and guess about how I reason or are you just going to simply ask me. "How do you reason.?.

You asked me to explain the reasoning of something that I couldn't accurately answer without asking the source. If you asked me "How does my dad reason?", then I would do the same thing. I would ask my dad for that answer. My answer wasn't replaced, you simply asked for an answer that someone else had to provide.

The thing that is scary to you isn't my lack of response. What's scary to you is how my Ai responded to that question. That is where your discomfort is and you expressed it by telling me that I need to Interact more with real people without understanding just how much I actually interact with real people.

You guys know me as "Jow Ga Wolf" the guy that does Jow Ga Kung fu, the guy that talks kung fu. So it's only normal for things to seem strange when I show a different side of me. I do things other than Kung Fu and me working with Ai is a must. If I want my job skills to be relevant in the future, then I need to be on the top of that game.

Part of that effort to stay on top of the game is to us Ai to build teaching tools and model how clear thinking works. Why is this important? Because without it, people will get Ambiguous outputs, misaligned reasoning, and mimicry (fancy word copying the behavior of the user.). All of that increases the risk of incorrect answers and harmful outcomes.

There have been cases where someone asked Ai to how to commit suicide and the Ai gave them a method and it worked. That's because of mimicry. So the last 7 months or so I've been learning about Ai and how to prevent Ai from doing things like that.

If there's one thing I've always been big on in Martial Talk. I don't want people to misunderstand who I am. I'm the same guy that talks kung fu but I also do other things.

 

[Prince_Alarming83 19]

*Darth Vader voice*
Your lack of response disturbs me

 

[HighKick]

*Darth Vader voice*
Your lack of response disturbs me

Well, like @JowGaWolf , I have other things going on in my life, this forum is a pea sized part of it. So, forgive me if I don't have the time to provide a fairly useless response when You think it should happen.
You seem to have an excess of time on your hands so clearly, we operate by a different clock.

 

[HighKick]

There's nothing straight-up scary about my lack of response.
1. You asked me to explain how does the Ai that I use reason. My question to you is "How would I know what Ai is computing or how it reasons without asking it?"

You want more human interaction? Let's do that.
2. Explain to me how I reason?

Now I wonder if you are going to sit there on the other side of that computer and guess about how I reason or are you just going to simply ask me. "How do you reason.?.

You asked me to explain the reasoning of something that I couldn't accurately answer without asking the source. If you asked me "How does my dad reason?", then I would do the same thing. I would ask my dad for that answer. My answer wasn't replaced, you simply asked for an answer that someone else had to provide.

The thing that is scary to you isn't my lack of response. What's scary to you is how my Ai responded to that question. That is where your discomfort is and you expressed it by telling me that I need to Interact more with real people without understanding just how much I actually interact with real people.

You guys know me as "Jow Ga Wolf" the guy that does Jow Ga Kung fu, the guy that talks kung fu. So it's only normal for things to seem strange when I show a different side of me. I do things other than Kung Fu and me working with Ai is a must. If I want my job skills to be relevant in the future, then I need to be on the top of that game.

Part of that effort to stay on top of the game is to us Ai to build teaching tools and model how clear thinking works. Why is this important? Because without it, people will get Ambiguous outputs, misaligned reasoning, and mimicry (fancy word copying the behavior of the user.). All of that increases the risk of incorrect answers and harmful outcomes.

There have been cases where someone asked Ai to how to commit suicide and the Ai gave them a method and it worked. That's because of mimicry. So the last 7 months or so I've been learning about Ai and how to prevent Ai from doing things like that.

If there's one thing I've always been big on in Martial Talk. I don't want people to misunderstand who I am. I'm the same guy that talks kung fu but I also do other things.

I doubt any response I give will be of any value to you, so I bow out.

 

[JowGaWolf]

The big problem I have with it is that it's very useful as long as you already know what you are looking for. So what's the point?

Yes and no. What you are referring to is the expertise that one must have to know when it's wrong. Sort of like, You may not know about Jow Ga kung fu, but If I gave you a punch combination that doesn't flow, then you can call me out on that based on your current knowledge and expertise.

This is what I would call "Ai configuration knowledge" I believe the correct term is Labled Data We configure Ai to be able to solve problems that we already know the answer to. That way we can tell if the configuration will yield correct resutls.

For me I call it Ai configuration knowlege becaue of this.

Here's how it would apply to Martial Arts.
Labeled data
Goal - Ai training = Teach Ai to match karate form to correct karate output. This means. Does my form match the input that was put into Ai of what my form looks like.
Me Configuring Ai
Goal - Teach humans (and Ai) how to digest martial arts form, movement, and applicatio clarity.

The difference here is that I don't want Ai to just look at the form. I want Ai to understand the form and to apply it's current knowledge of biomechancis so that it can determine if the form that I'm doing is correct for learning the application of the technique. I do not want to know if it's correct to be a copy. I want to know if it's correct for application.

Why it matters? I can load this combination into Ai and tell Ai that it's correct Form. But if I asked my Ai is this correct application or practical use then it needs to be able to figure that out. It will tell you that you or doing correct form. But will it tell you that it's practical.

I used the picture above to highlight the difference. If I tell Ai this is correct form then it will look at my body postion to see if my body position matches the input that was said to be correct.

Here's how my Ai responded.
If you block low with your left hand and then punch with the same hand, it can work—but only in certain situations. Let’s break it down by how the body moves, how fighters behave, and what the move is meant to do.

Keep in mind I did not ask in which context this combination is done. I didn't say if it was against a punch or kick. It's gotta figure that out on it's own. While normally I wouldn't do this, I wanted to see if it could figure it out. I only provided the image above as a reference.

Me:
Can you create an image that represents this "Fighting Behavior Context Practical Use Case: Works best in close-range trapping, sticky hand terrain, or low-line feint and follow-up" I'm having a difficult time visualyzing it. Make the picture look realistic and follow the biomechanics of humans. The person that will do the low block should have a green belt.

So the guy on the right is the one doing the low block. I had it alter the color of the belts so I could tell what Ai is visualizing.
Ai:

So this arm placement doesn't make sense to me. There are ways in which , low block and middle punch work with the same hand , but it doesn't look like this, and the block isn't like the one shown in the TKD picture. In the one I know the block is more like a redirect / scoop away. So I want to know what Ai is looking at.. So I asked it to give me a walkthrough.

Ai.

Step-by-Step Breakdown: Left-Hand Low Block → Left-Hand Punch​

Context: Close-range sticky hand terrain, green belt practitioner (left side), realistic biomechanics

1. Engagement Phase

• Both practitioners are in close proximity, hands raised in monitoring contact (sticky hands).
• The green belt has left hand forward, right hand chambered near the ribs.
• Opponent has right hand forward, preparing a low attack.

Me: I don't waste time trying to imagine what this looks like, becasue I need to understand what Ai is seeing. So this is what Ai is seeing

Here’s your image, Marvin—rendered with realistic human biomechanics and tuned to your diagnostic terrain. It captures the engagement phase of the sequence:

• The green belt practitioner stands with left hand forward, right hand chambered, and knees slightly bent
• The opponent is in close proximity with right hand forward, preparing a low-line attack
• Both maintain sticky hand contact, threading centerline control and compressed stance

Me: When it comes to martial arts I already know there is an issue with identifying Right and Left which for us would be a big issue. So asked what type of attack was the opponent doing? And it showed me this.

Ai:
In the image, the opponent is executing a right-hand low punch aimed at the green belt’s left-side ribs or abdomen. This type of punch is often used in close-range terrain to target the lower torso, bypassing high guards and exploiting compressed stance openings.

Diagnostic Breakdown of the Attack​

• Arm Used: Right hand
• Target Zone: Left-side ribs, floating ribs, or lower abdomen
• Biomechanics:
  
  • The punch is delivered from a slightly bent stance
  • The torso leans forward slightly to compress and drive the strike
  • The left hand remains chambered, preserving centerline balance
• Tactical Purpose:
  
  • Disrupt the opponent’s base or breathing
  • Feint low to draw a reaction and open up high-line counters
  • Thread into sticky hand trapping if the punch is intercepted

So that's what Ai is showing and here is where your current knowledge becomes important. Does this combination work as Ai described it? Keep in mind of the left and right issue that Ai has specifically in CoPilot. I don't know if Chat GPT has that same issue..

The only way you know for sure if this is posible is if you have experience. If you see something wrong, then you need to correct Ai based on what you know in on what your martial arts system teaches. Ai will then use that information to make better informed Analysis about martial arts.

This is why I say that the problem is that Ai doesn't know how to do martial arts. The one's that some nerd Ai programer will teach will be the popular martial arts like Karate and TKD where there are exmples of people using the actual techniques found in the forms. For the rest of is. If there's no video of people using the techniques that are actually trained then your system won't be included because there is no application reference for Ai. Unless they use the Demo stuff, You would need to continue to do this type of configuration for every combo.

What am I visualizing vs what Ai is analyzing? Is Ai seeing what you are seeing? The only people who are qualified to do this type of configuration are those who know how to use martial arts.

This is the process for a lot of Ai "training" from the user end. Before you ask Ai how to do something, you better make sure you and Ai are on the same page as to "What needs to be done." Some things are easier to get a reliabl eanswer but everything has to configured.

Vs labeled data which basically says, you body should follow this body position from the form.

 

[JowGaWolf]

The big problem I have with it is that it's very useful as long as you already know what you are looking for. So what's the point?

By the way. Lots of technical stuff is usless until you configure it for your use. Hence the setup for smart TVs. It doesn't work until you set it up, and part of that set up is so that companies can capture your viewing habits.

 

[philwhite]

The big problem I have with it is that it's very useful as long as you already know what you are looking for. So what's the point?

I think this kind of harkens back to dead internet theory in that search engines used to return results that were useful (and human created). Now they only return results from a specific set of sites or ones that have paid ad revenue. Your only choice to get information now is unfortunately through their AI which they try to get you to pay for. Its a sad reality. Its like locking real knowledge behind a paywall or worse selectively ignoring the truth.

I don't think we are required to be joined at the hip with AI. Unfortunately the workforce in the corporate world is heading that way since stock numbers are tied to AI adoption and AI job reduction. In truth I find that most people who use it tend to expose thier ignorance when doing so as they believe the answer too deeply or lack the knowledge to challenge what is returned.

I just brought the initial topic up since I've been querying some things that I could not find answers for using search engines, yet AI, whether right or wrong produced some decent responses.

 

[Prince_Alarming83 19]

Well, like @JowGaWolf , I have other things going on in my life, this forum is a pea sized part of it. So, forgive me if I don't have the time to provide a fairly useless response when You think it should happen.
You seem to have an excess of time on your hands so clearly, we operate by a different clock.

Dude, it was directed at him. And a joke. Sorry if it caused offense.