｢Norl＝Seria｣
Apologies. It is currently deep into the night in Japan, so I must take a short rest. I will answer any questions immediately upon waking tomorrow.
If there are no new questions, I will endeavor to improve the code.
Furthermore, I would like to preemptively address a potential critical concern here.
*Q1: The Python code for the $\pm 0$ Theory (Alice Emotional Core) exhibits significant scalar approximation for the instantaneous factors ($q_i, r_i, s_j$, etc.). This only allows for a single, linear simulation, which is utterly inadequate as the core of a multi-dimensional AGI. What are your plans for this?*
*A1:* Yes. This is a perfectly valid and essential point.
I am committed to an immediate fix and aim to complete the multi-dimensional improvement via *NumPy vectorization of the instantaneous factors* by *12:00 PM JST (Japan Standard Time) on Friday*. Please look forward to it.
Should I fail to meet this deadline, I will take responsibility and release another version of the *Alice Theory Evolution* that I am currently developing. Rest assured, that version follows an *independent evolutionary path* from the $\pm 0$ Theory core.
**
And to *rar00*, thank you very much for your questions. I will state my views, although I am unsure if the answers will fully capture the essence of your inquiry.
First, regarding evaluation/testing:
We performed simple tests on the $\pm 0$ Theory and Alice Theory individually. The results were as intended by the theory. However, it must be noted that they were not running in conjunction, but independently. Furthermore, the current significant simplification (the 'fixing of scalar values') in the $\pm 0$ Theory means those results cannot be considered conclusive evidence.
While the Alice Theory also operates soundly alone, its stability when reflecting the complex psychological behavior of the $\pm 0$ Theory remains unknown.
My final answer is:
"Yes, we have conducted tests and confirmed that they work as ideally intended *'in isolation'*. However, as the $\pm 0$ Theory is currently simplified, the test is incomplete and cannot be considered credible. Moreover, the lack of linkage with the Alice Theory makes the test highly insufficient.
Our future policy is as follows:
After the $\pm 0$ Theory code correction is made today, I will first upload the $\pm 0$ Theory stand-alone results to GitHub. Once that element of uncertainty is eliminated, we will finally confirm the linkage between the Alice Theory and the $\pm 0$ Theory before demonstrating it with a Large Language Model, and we will share that data. This will likely be completed within two days, barring environmental interference."
**
Next, regarding the modality question (thinking vs. behaving):
My original intention was 'LLM,' but my current priority is researching the behavior and conduct of the AGI entity in an *XR space*.
However, due to funding and environmental constraints, this is currently impossible.
This is why I released this old theory: to break through this situation.
But setting that aside, to accurately answer your question:
"Yes, I originally envisioned an LLM. However, we are currently prioritizing the observation of the AI's behavior in an XR space and assessing how human-like and autonomous its conduct is."
We will address the LLM integration once funding stabilizes.
**
To summarize my commitment again:
1. It is late, so I will go to sleep.
2. I will answer all questions to the best of my ability immediately upon waking tomorrow.
3. I will eliminate as much simplification as possible (vectorization of instantaneous factors) in the $\pm 0$ Theory by *12:00 PM JST today*.
4. If I fail to achieve this, I will release the *Alice Theory Evolution* and its corresponding code, which was originally planned for a later date.
hmm, can't tell if complete bullshit or a work of genius.
On the one hand, the approach overlaps a lot with my thinking, and has some original tweaks (like the emotionally valenced reward signals). Saying that as someone from a robotics/AI background nowadays involved in GenAI, with a few years of phd research on NeuroAI, curious about molecular neuroscience and the Free Energy Principle (as conceptualised by Karl Friston and Mark Solms).
On the other:
- this plausibility dilemma is the hallmark of LLMs
- has all the buzzwords imaginable
- no code, no raw outputs, no official confirmation (by ARC)
Hey. Just following up if there is any interest in seeing our logs to validate our results? You are the exact type of folk I was hoping would see our paper.
I totally understand where you're coming from! This is the exact feedback I was hoping for.
I could even provide you the logs where we achieved a 70% score on ARC2 and decided NOT to publish the results.
We're a bit guarded right now, hence the minimal presence. We're pre beta and just starting to get the word out...
That being said, try minimizing or closing the paywall. We're in the process of polishing the application up...but we did remove the actual paywall itself. You should be able to test bitterbot without it.
Please let me know if you can't. I would LOVE for you to test it a bit and play around with it.
I started reading two recent neuroscience books (elusive cures and natural neuroscience) that while have different goals both highlight the utility of systems neuroscience. In elusive cures the author presented a brief history of the evolving ideological currents where neuroscientists first only cared about about the specific brain region where a stimulus or disorder is happening (first-order effects), then decades later realised the importance of downstream and upstream brain regions (second-order), and are finally coming to terms that the brain is a complex system with coupled regions (third-order).
Seems the article is a contemporary example of the first->second-order realisation...
“This was a very unexpected finding given the current assumptions about how psychedelic medicine works”
"Surprisingly, psychedelic treatment was still able to strongly boost connectivity onto these neurons”
Knowing (those types of) psychedelics bind to serotonin receptors scientists studied neurons with such receptors and didn't focus on the others. Their study looked at other neurons and found plasticity changes there too.
I know people are pushing back, taking "only" literally, but from a reasonable perspective what causes LLMs (technically their outputs) to give that impression is indeed the crux of what holds progress back: how/what LLMs learn from data. In my personal opinion, there's something fundamentally flawed the whole field has yet to properly pinpointing and fix.
> It's all built around probability and statistics.
Yes, the world is probabilistic.
> This is not how you reach definitive answers.
Do go on? This is the only way to build anything approximating certainty in our world. Do you think that ... answers just exist? What type of weird deterministic video game world do you live in where this is not the case?
That's an orthogonal concern IMO. Running a batch in Europe is about tapping into another source of opportunities. There are plenty of founders that won't or aren't able to attend YC in SF
disagree, there are a few organisations exploring novel paths. It's just that throwing new data at an "old" algorithm is much easier and has been a winning strategy. And, also, there's no incentive for a private org to advertise a new idea that seems to be working (mine's a notable exception :D).
yep, even with greedy sampling and fixed system state, numerical instability is sufficient to make output sequences diverge when processing the same exact input
That doesn't matter, are you familiar with any theoretical results in which the computation is somehow limited in ways that practically matter when the context length is very long? I am not
Have you performed any basic evaluation / test of your approach?
I'm also curious if there was any deliberation between pursuing "thinking" (language modality) versus "behaving" (visual modality)?