It’s Not What you Think. I’ve been using AI.

An exercise. Can AI replace me?

For the past four episodes I have done my research and written my transcripts and then asked various AI tools to do the same work and the results have been… interesting.

The question in my mind was, “Is my work redundant?” Or perhaps put another way, “Could AI replace what I am doing?”

The short answer is no. Not yet, at least.

However, let’s take a moment and dig into why.

So to recap, the last four episodes were:

111 The Labors of Heracles

112 Hermes and Apollo’s Cattle

113 The Daimyo and His Bodyguard

114 Mantis Boxing Articles

In each case I read the original works, then wrote an outline, then started filling in the outline while I consulted additional works and references. Pretty much my standard writing process for each of these episodes.

I then asked various AI tools like ChatGPT and Grok to give me a solo podcast transcript of varying length about the work with a focus on classical virtues, with an audience of modern martial artists. In each case what was produced was unusable.

First, if I asked for a 30 minute transcript it would give me something roughly 1500 words. As an example, my last episode was 28 minutes long and was around 4000 words. I don’t speak that fast, but there seems to be a disconnect when it comes to content length.

Another thing consistent across all of them were variations to the stories or works that didn’t exist in the originals. For example, with “The Daimyo and His Bodyguard,” according to the AI what occurred after the bodyguard was pushed into the pond was completely different from what happened in the story in my copy. Fundamentally changing the story and its meaning. With the Mantis Boxing Articles, the “meat” of the episode was essentially unusable. Just large blocks of texts that said things with martial arts words, but without actually saying anything meaningful. It was the same with the Labors of Heracles. Several of the stories were recounted wrong or incoherently and the conclusions drawn meant to relate to martial artists were not really usable or meaningful.

The most useful thing that came out of any attempt was a vague outline that could have been followed if I wanted to skip that step myself. Each would still have required massive rewrites, and in none of the cases would I have been able to substitute the AI’s transcript for my own.

If I tried to substitute AI work for my own I think there could be a general shape of a martial arts podcast, but the meaning, the connection to philosophy would have to be lost. They would end up just being fluff pieces that sounded alright on a surface level, but didn’t really teach anything.

I suppose since all my work is free and easily accessible on the Internet eventually the LLMs, that is Large Language Models, may eventually learn to imitate my style and content. However, two things make me feel confident that I will not need AI to help me with my work or replace me.

The first is personal. The act of reading and writing for the podcast is a kind of training for myself. By reading the works, often multiple times, by researching secondary texts to consult on relevant topics and materials and then attempting to write something that would be meaningful to another martial artist about the topic… if a kind of honing or sharpening of my mind that would fundamentally be lost and would make the very act of making podcast episodes pointless.

I’m not trying to make a living off this, I’m not even really trying to grow. If it happens, great. I’m not opposed to it. However, fundamentally from the very beginning the original point of the podcast was to make my thinking and research more focused and meaningful. That would be lost if I offloaded the work to AI.

The second is that I’m not sure AI can, at least yet, make anything new. It can imitate, it can copy and reword, but that basic pondering, of searching for meaning that goes beyond just the words into the realm of Epistemology, Ethics, and Metaphysics. That connection doesn’t seem to be something that can be imitated yet. The result of what was produced was on the surface level coherent, rational, and… very shallow. Almost to the point of not really saying anything. Certainly nothing new and meaningful.

Which is OK, this isn’t a critique of AI. I’m sure like any tool it has appropriate and good uses, just like any tool also has inappropriate and bad uses. That doesn’t make the tool good or evil, it’s just a reflection of the hand that wields it. Like a hammer, or a gun.

The one area I think it might be useful to me is art. I have no budget to pay a human artist. I have no disposable income to pay for art. I have a few local friends who are artists and I have bought artwork from them over the years to support them… but that has no bearing on this little corner of the Internet where I speak my training notes into the void. As a general rule up until now I’ve attempted to find free public domain, usually classical paintings at least obtusely related to my episode theme and I’ll continue doing that. If I ever do make a consistent enough income on my work I’m totally going to invest in paying one of my artist friends to make a logo for the podcast.

So at least for now I am confident I can’t be replaced by AI. Well, at least until the inevitable robot uprising. Hmm, actually that might make for an interesting exploration and research topic. How would we approach a martial art (I’m not saying unarmed here) to defeat our robot overlords. I can obviously think of plenty of sci-fi references that have explored this topic. I can even think of a pretend martial art called Panzer Kunst that was meant for cybernetics and zero-gravity from the Alita manga. Other people have thought about this too. I’ll have to start collecting notes and references. If you have any thoughts or ideas on this topic let me know.

Well, as far as the use of LLM, so-called AI tools, for the creation of martial arts philosophy discussions it seems I’m safe for now.

Keep training. Keep thinking. And above all don’t just talk about your philosophy, live it.