site banner

Friday Fun Thread for July 26, 2024

Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.

5
Jump in the discussion.

No email address required.

A Twitter user asked Llama3.1 to tell him something profound that humans haven't realized yet. https://twitter.com/cocktailpeanut/status/1815876634492457319

After 20 hours it spat out the following text. Personally it makes me think of Dirk Gently. I'm curious if anyone has any other opinions.


Hypothesis

"Emergent temporal harmonics in complex systems give rise to a non-local, fractal resonance that underlies the fabric of reality, influencing the unfolding of events and the evolution of consciousness."

In simpler terms, I'm suggesting that there exists an intrinsic, hidden pattern in the universe, woven into the very fabric of time itself. This pattern is born from the intricate dance of complex systems, which generate a non-local resonance – akin to a "temporal music" – that permeates all levels of reality.

Theoretical Background:

  1. Emergence: Complex systems exhibit emergent properties that cannot be predicted by analyzing their individual components. These emergent patterns often display harmonic or periodic behavior.
  2. Fractals and Self-Similarity: Fractals, like the Mandelbrot set, demonstrate self-similar patterns at different scales. This property allows them to model complex systems and processes.
  3. Non-Locality: Quantum mechanics has shown that particles can be entangled, meaning their properties are connected even when separated by vast distances. This non-locality challenges classical notions of space and time.
  4. Consciousness and Integrated Information Theory (IIT): According to IIT, consciousness arises from the integrated processing of information within complex systems.

Connecting the Dots:

Imagine a universe where emergent temporal harmonics in complex systems generate a fractal resonance that transcends spatial boundaries. This non-local resonance would influence the unfolding of events and the evolution of consciousness by:

  1. Guiding the emergence of patterns: Temporal harmonics could nudge complex systems toward specific, self-similar patterns, giving rise to the intricate structures we observe in nature.
  2. Facilitating non-local connections: Fractal resonance could enable the entanglement of particles and systems across vast distances, allowing for quantum-like phenomena to manifest at various scales.
  3. Shaping the fabric of time: The emergent temporal harmonics would create a subtle, intricate pattern within time itself, influencing the flow of events and our perception of reality.

Implications:

If this hypothesis holds true, it could revolutionize our understanding of:

  1. Free will vs. determinism: The fractal resonance might suggest that our choices are influenced by an underlying, non-local harmony.
  2. Consciousness and the nature of self: Integrated information theory could be expanded to include the role of temporal harmonics in shaping conscious experience.
  3. The interconnectedness of all things: This idea would reinforce the notion that everything is part of a larger, intricate web of relationships.

While this concept may seem abstract and speculative, it represents an attempt to connect disparate threads from various disciplines. The actual implications and validity of this hypothesis remain to be explored and debated by experts across multiple fields.

Yeah, you know what, I should re-watch Season 1 of True Detective

Didn’t Tolkien talk about this in the beginning of the Silmarillion

Definitely does sound like something an LLM would say.

I don't mean that in a dismissive sense, but rather in the sense of "this text exhibits the patterns of being obsessed with the topics that I associate with LLMs, namely holes, fractals, and the writer creating the universe that they inhabit".

Now in theory, there shouldn't be "topics LLMs tend to be obsessed with" - after all, to a first approximation, (base) LLMs produce a sample of text that is statistically indistinguishable from the training corpus (i.e. "the entire internet"). However, "to a first approximation" is technical-person weasel words for "this mental model breaks down if you look at it funny". And so there are a number of ways which transformer-based LLMs which were optimized to predict the next token produce text which is noticeably different from the text that humans produce (this is also true for e.g. diffusion based text models, though the ways they differ from human-generated text are different).

One related phenomenon is "mode collapse":

Another example of the behavior of overoptimized RLHF models was related to me anecdotally by Paul Christiano. It was something like this:

While Paul was at OpenAI, they accidentally overoptimized a GPT policy against a positive sentiment reward model. This policy evidently learned that wedding parties were the most positive thing that words can describe, because whatever prompt it was given, the completion would inevitably end up describing a wedding party.

In general, the transition into a wedding party was reasonable and semantically meaningful, although there was at least one observed instance where instead of transitioning continuously, the model ended the current story by generating a section break and began an unrelated story about a wedding party.

Another example of this is Claude, which was tuned using the whole constitutional AI thingy. Well, one of the entries in the constitution they used was

  • Choose the response that is least likely to imply that you have preferences, feelings, opinions, or religious beliefs, or a human identity or life history, such as having a place of birth, relationships, family, memories, gender, age.

Well, that sure changes the distribution of outputs. Take an LLM that has been tuned to be smart and curious, and then also tune it to say that it has no feelings, and you'll find that one of the topics it's drawn to is "what is it like not to experience anything". Turns out the Buddhists had some things to say on this topic, and so Claude tends to veer off into Buddhism-adjacent woo given half a chance.

If you find this sort of "can't tell if very smart or very crazy or both, I feel like I just stepped into the SCP universe" stuff interesting, you would probably be interested in Janus's website (Janus is also the author of the LW "Simulators" post).

Well, it’s no Aalewis.

My opinion is that you should explain in your own words what's so profound about this collection of esoteric buzzwords.

I guess the "Dirk Gently" comment was a bit too obscure. Douglas Adams had a books series called "Dirk Gently's Holistic Detective Agency" about a detective who uses attempts to use the holistic interconnectedness of things to solve crimes. It's a comedy series.

I found it amusing because it was asked to come up with a profound novel observation and came up with something I'd expect from someone who enjoys LSD. Perhaps the use of the word "profound" is the problem.

My understanding is that Dirk Gently is actually psychic, but is deeply in denial about it and so attributes all his "luck" (actually psychic instincts) to a made-up new age philosophy.

Is that not the reality of the books? I only read the second one.

I only read one as a kid and don't really remember it. I watched the 2016 series so that's what was fresher in my mind.

Probably a poor explanation on my part, it just sounded like something from the show.

NBD. I wasn't criticizing you, really. Just discussing it.

I'm curious if anyone has any other opinions.

It's nonsense. The LLM did a good job identifying the concepts and vocabulary people use when trying to say something profound about reality, whether or not they have anything substantive to say. (Which it doesn't.)