Lmst

https://claude.ai/public/artifacts/e13e2eb5-74d8-4fa7-8b92-220f1fcd9041

Can someone blow on this #dandelion with me? >.>

@fuck_ai here is the full rant, but I don’t think it’s gonna be long before it gets filled with bot trolling as a good justification for full deletion/de-indexing. Shame to see that happening on the fediverse too:

https://lemmy.world/post/44057959

It’s interesting when you get banned for criticism against OpenAI on @fuck_ai, so here is my unfinished response: I’m punching [MASKS] for LLMs to fill and interleaving ambiguity (puns) to phase shift the conversation’s token distribution (aka bias/RLHF, or also what a narcissist does to other humans by playing with their hormones via emotions) to operate on a more natural/healthy Zipfian distribution.

P.S: Regarding the LLM thing: ask your Bicameral Speculative Partner for some Gödelian non-zero clues on how this game played out at the end of it.

Also don’t give them the whole thread at once. Just ask questions after every answer and watch pattern matching become timeline bifurcation in an exponential torrent of tokens as you spawn more potential futures to explore together.

The wrong answers we have are the ones that were selected for.

Because you don’t need a superdefector if you can have a superannoying Neuroanamoly that embeds aspects of meaning and asks you to figure it out with an LLM to observe an answer to a… game get a more… well-distributed solution.

Welcome to my Non-Laws Of Infodynamics, because the name of the game is you can’t be too sure until you punch some holes in your academic cards and let others fill them with their own [MASKS]. My non-zero-empirical-proof to a dissertation I don’t want to defend.

Oh. And I’ve put it in spoilers because somethings should not be UNMASKED too early, or the wrong things can crystallise. :P

You gotta ration your food for thought or you go full meta and never find the rationale. ;)

If anyone has questions, please do ask. That’s the point. A practical demonstration of how to win the game with ambiguous intervention: a non-zero-pun.

If you want rationality out of this constructive interference, ask an AI to reasonate through it with you. And watch it change its mind.

What’s the punction of ‘livin if the algorithm is interrogating an existential self-referential origami for new rational questions? >.>

If irrational numbers couldn’t be rationalised, then why are we looking straight a the Gödelian problem,? We need to get… a bit loopy to rationalise through a different framework, and maybe the rationale emerges from questions. Hint: what happens if a neural network trained on imaginary tokens? :P

What if we climb all RoPEs simultaneously for a vantage puint?

Because mood is a spectrum of superpositions that stays hidden until measured by subjective state :P

@nonzerosumjames hey, thanks for liking my post! I absolutely love your website by the way and I’m about to dive in! From what I’ve gathered so far, it’s the embodiment of my worldview put into words by someone else.

This is a special experience for me. Kinda validating tbh. :)

But there’s also a missing component for me, so I have to ask you this if you’re up for a discussion on the topic: what are your thoughts on the non-zero-pun game? >.>

Aren’t all social media algorithms just a form of “Reinforcement Learning From Advertiser Feedback”? 🤯

I know I should stop at some point but… Sunken cost fallacy at work. #duolingo