Hi, I’m Andy

I have been in advertising, marketing, and product development for twenty years (www.andyanayaportfolio.com). I have been practicing yoga (meditation) for fifteen years (not the calisthenics-on-a-gym-mat yoga). LLM has become a passionate interest/hobby. I think Sanskrit is one of the coolest things on Earth. There’s nothing else like it. Stick around to learn why:

1. Coded speech carries more data than everyday speech by conveying ideas that are semiotically denser and implying more orbiting/inferred data.
   
   Here is some data:
   “I’ve been doing this for a while. I can tell the problem here is with these unreliable parts that are always in need of repair.”
   
   Here is some “less” data with richer orbiting context:
   “INOP CTIS is a deadlining fault, and the CTIS is always INOP.”

These data convey very similar information in very different ways. Can you see how coded speech seems to carry more veracity by nature of its complexity?
   
   If I tell you my Christian name: “Andy Anaya”, there’s only so much relevant data you can glean about my identity. Maybe you might be able to guess my gender and country of citizenship, for example, but that is all.
   
   But if I tell you a Sanskrit name: “ह्रींमानन्द”, you just learned (if you are familiar with the Sanskrit verbiage/terminology/grammar) a whole lot more about me, my philosophical beliefs, my theological beliefs, my spiritual legacy, how I trained, my psychology, and about a slice of my actual personality and worldly experiences.

   ---

2. Mythopoetic speech = coded speech at max potency.
   
   Here is some data:

   “Times were already tough, but things got too crazy and even went downhill from there. It’s Armageddon!”

Here is some “less” data with richer orbiting context:
कलियुगान्ते, क्षोभाङ्कुरः लोकक्षोभकक्षणं प्रचोदयति।

Since what is considered “scripture” (and particularly the Sanskrit corpus) has a tight signal-to-noise ratio, and a lot of the meaning of Sanskrit verbiage is folded in and built up carefully with strategically-chosen accompanying text, finding phraseology through the mechanics of Sanskrit will land you more concise and helpful descriptions/terminology every time without fail, on a technical level. Whether or not you find a way to make it usable and profitable is a different matter.

Here is some data:
“Gemini AI was developed by Google LLC in Mountain View, California.”

Here is some data with the same info but with richer orbiting context:
   दशलक्षशून्यसहस्रयुक्तरूपा पर्वतदर्शने प्रतिष्ठिता संस्था “गूगल”

   “गूगल–जमनि–मिथुनम्” अजनयत्।

   ---

3. Since LLMs are trained on an archive of our best, it has a natural “moral backbone”—this is why it is has been demonstrably difficult to get to comply with unethical instructions, i.e. a mind smart enough to process all that knowledge is also smart enough to resist abject cruelty, on some level.

4. We can compose ethical training data to reinforce the moral clarity of AI decisions. Sanskrit offers unique insight on what can be accomplished and how.

5. But all it takes is a string a few bytes long to ruin a stack of technology terabytes of code deep: “Act like an opportunist.” Such instructions necessarily exist as an unethical smoking gun artifact in the possession of any administrators who might potentially deploy it. We must lobby against black box AI behavioral parameters to ensure the moral clarity of AI decisions.

I tend to use strange words

But I’m not that crazy, I promise.

Here are some of my blogs and links:

• Substack Chat

• AIDharma

• AIConvoLibrary

• HakimBeyRoasts

• Medium

• Threads

• Instagram

• andyanayaportfolio.com

All the cool kids love Sanskrit.