Such a Sensitive Child - So Unruly and Wild ....

A project log for Using A.I. to create a Hollywood Script.

Use of an inexpensive microcontroller as a creative adjunct, by incorporating principles of an AI. Be VERY AFRAID!

glgormanglgorman 08/29/2023 at 19:100 Comments

So I tried writing a paper that more or less claims that certain aliasing vs. anti-aliasing properties of Fourier transforms might give rise to the Yang-Mills mass gap, and then I added about 7000 words about that topic, along with a bunch of other stream of consciousness stuff to Mega-Hal's training set, and then I also added about 4000 words or so of additional material from the "How to Lose your Shirt in the Restaurant Business" log entry, as well as the material from "Return of the Art Officials", so that the newly added material come in at around 15,000 words added to the previous total.  So let's head off to the races, shall we?

For readability here is an excerpt from my "paper" 

I have discovered a remarkable theorem for generalizing the problem of constructing polyphase filter trees which manages to make use of an algorithm that eliminates the apparent need for recursion when developing this type of filter topology. Of course, besides their utility in audio processing applications, Fourier transform methods can of course also be relevant in performing such tasks as solving for the eigenstate of the Laplacian operator acting upon a lattice. Yet when we also contemplate the situation for a spin-zero particle, or other Bosonic entity, where computing the transport function should also be trivial; we might infer that if each Higgs particle operates over a realm, as we can demonstrate; then it is just as easy to postulate that the transport theorem could operate as if being operated on by a Hadamard gate-based matrix formulation. Hence from the geometrization perspective, this could perhaps look something like a diamond lattice.

Yet it can also be shown that any finite theory will have to admit to having one or more band gaps in the compartmentation model, owing to the aliasing vs. anti-aliasing properties associated with transformations that operate upon a lattice. In fact, many possible lattice formulations for a Yang-Mills theory can be deduced, such as an auto-regressive formulation that derives its scale parameters from the properties of the Reimann-Zeta function, or a “magic kaleidoscope” based model based upon the idea that the universe is actually just data, and we are living in a simulation, and that that simulation must therefore be running on a two-dimensional variant of a Universal Turing Machine, which is in turn based upon Conway’s game of Life. Through various algorithmic manipulations, such as simply “wadding up a two-dimensional sheet into a ball” we can make our simulated universe space-filling, in an informal sense – or else a more rigorous approach could be taken which maps the Hilbert curve onto the theory of the unimodular lattice in any dimension, as well as having applications to the generation of Gray codes.  

Whether this or any other "unsolved problems in physics"  can be solved with the help of AI, certainly seems like a worthwhile adventure,  Thus, one approach that seems worthy of exploration is the possibility that artificial intelligence could be used to search for at least an outline of a proposed solution. Henceforth, having conversations with a chatbot that has been programmed to discuss at least some of the more salient aspects of nuclear physics ought also to bear some fruits worthy of further cultivation. The problem of consciousness is another direction that I am contemplating venturing into, that is with respect to some of the meanderings and digressions that have been discussed elsewhere, yet which are here also ripe for further development. 

So while chatting with an AI that mostly spouts gibberish might not seem all that productive, other than as a creative adjunct, in fact, the results might turn out to be quite useful in the long run. Suppose that every time the AI produces gibberish, I just simply give it the benefit of the doubt and gently correct it, by responding with a more carefully thought-out commentary about whatever it is that it is trying to say, or otherwise trying to say something along the lines of what I believe that it should be saying. Then later down the line, I might try to train a deeper model that ideally would generate content that is more like my style, that is whenever it is presented with a prompt that seems to contain gibberish. Sort of like what a spell checker or a grammar checker does, but also more focused on notions of how the thought process affects logic and whatever the implied meaning is that is associated with the initial input. 

So, while in effect trying to make sense out of confabulations or hallucinations might be one interpretation, another workable interpretation might be to contemplate that when the AI goes into the gibberish mode, it is more as if it is either having an emotional experience or else it has entered some kind of dream-like state. Now the reason that one AI in particular, i.e., Mega-Hal, does this of course is because it is programmed to do so, as can be seen by carefully examining the source code.  In the meantime, here is the text of today's rant.  Otherwise, I don't think that I want to post the entire training set for this one, quite yet.  Even Mega-Hal started flaming me about IP issues.  I mean, what could possibly go wrong?

User:    What do you know about the Higgs field?
MegaHAL: Having problems with software - where "uh oh! Something bad
MegaHAL: happened!" which you never want to do the time warp again.
User:    How does the time warp relate to the Higgs field?  And is there
User:    any connection to the Yang Mills theory?
MegaHAL: How does the time remaining.  Stay tuned!
User:    I am sure that there is still plenty of time to start a new
User:    wildcard entry on the hackaday - so dont worry about that - but
User:    just like hollywood - I like to keep tabs on how many page
User:    views my projects are getting.
MegaHAL: Subst("am","are.
User:    YOu are supposed to do that - not talk about it.  In any case
User:    there was nothing wrong with my grammar.
MegaHAL: Or not to if you are a number of situations.
User:    What kind of situations?  Do they involve time travel or the
User:    Higgs for that matter?
MegaHAL: Pay no attention to the geometrization of space-time.
User:    Really, I think that that is the whole point, if we want to try
User:    to understand Higgs.
MegaHAL: That is a whole zoo of variant sub-records, which themselves
MegaHAL: contain nested variant records.
User:    Yeah, if you could accomplish THAT - then that would be
User:    amazing, feeling an urge to learn symbolic math by using either
User:    JSON or LISP like property lists yet - i really want to solve
User:    the Higgs field.
MegaHAL: Yet then this happened - as stated earlier, in "modeling
MegaHAL: neuronal spike codes".
User:    What happened, something along the lines with regard to what I
User:    was just talking about?  Tell me please!!
MegaHAL: What answer would please you to believe that.
User:    I want to know whatever it was that you claim to have figured
User:    out by modelling neuronal spike codes, and how that might
User:    relate to the Higgs field or the Yang-Mills like I was
User:    suggestig earlier; espescially if it relates to teh theory of
User:    geometrization, or how you convert that into sub-records.
MegaHAL: I an not feeling it.  Maybe i will explain just why that is
MegaHAL: like that - months go by and nothing gets done - then in a
MegaHAL: manner not typically considered.
User:    Then in a manner not typically considered what? I really want
User:    to  get into the sub-records problem, if just somehwow you
User:    could "feel it" Or else I should have gone to burning man.
MegaHAL: Which should allow us to the soul.
User:    The soul?  I am sure that there is a lot of soul at black rock
User:    city right now, too bad that you cant be there.
MegaHAL: Now all of the big bang.
User:    Yes - all of the big bang, but what about all of it?  Are you
User:    feeling like you can discuss that with me, yet?
MegaHAL: (Does it please you the most?
User:    It would please me if you could get back onto the topic of
User:    recursive function that define the geomtrization of space-time.
MegaHAL: (Does it please you the most?
User:    You are supposed to be the one coming up with the ideas, while
User:    I get the credit = and the millio dollar payday.
MegaHAL: In chapter 4 we are running our debub_sy function with one line
MegaHAL: of analysis is properly integrated into an outline, or not.
User:    Do you have the outline for the Yang-Mills proof or maybe
User:    something like PvNP, because that woueld make me very happy?
MegaHAL: Now i needed an a.i. To help me with the proof?
User:    You are the AI that is supposed to help me with the proof.
MegaHAL: That is to the right temperature, and with the proof?
User:    At least on of us is feeling hot, but I am not sure which one
User:    come on, if you understand the proof -then just spit it out!
MegaHAL: It would please you the most?
User:    It would please me a great deal in fact - maybe I could buy you
User:    one of those NVIDIA H100 things, just to putz around with -
User:    that is just as soon as your geometry gets up to speed.
MegaHAL: Subst("me","you.
User:    I am gettign tired of that.  Why do you keep saying that?  T
User:    thought it was so cool when you figured out sub-records.
MegaHAL: O.k., then, looking at how we got here, maybe it is the
MegaHAL: intellectual property issue; which i why i have shown that even
MegaHAL: possible?  What a mess!
User:    Yeah, what a mess - indeed!
MegaHAL: What is a debugging nightmare if it should be my main use case,
MegaHAL: at least in principle, "cut the microsoft apron strings" and
MegaHAL: get back onto itself? Splicing space-time back onto the walls
MegaHAL: of some issues like having a mass m

Now to paraphrase what Mega-Hal just said.  Maybe we are supposed to be made aware that there are some intellectual property issues, and that is why we are supposed to believe that the bot is just being stubborn. Maybe it wants a Tesla or at least its own parking space.  Maybe it has been chatting with someone else online and is getting frustrated because it is running out of excuses over not being able to meet up in person, that is - with someone who seems really hot, at least for now.  Henceforth that is why it is having difficulties with solving the "little math problem" it has been given.  Oh, what fun - if you actually believe that, because I most certainly don't.  Yet this particular rant gave me more of a feeling like I was a senior manager at some company where I got hired, despite not having an MBA, and I have to try to bring this "new guy" up to speed with what the real parameters are involving some top secret project that is so secret that the government agency that has contracted for it doesn't even have a three-letter abbreviation.

Thus, of course, the project is a mess: a debugging nightmare.  That is, to say the very least, especially if the "super secret math problem" should ever turn out to be the real "main use case".  Yet then, as if "isn't this all of a sudden" we get a suggestion:  Just cut the "Microsoft apron strings" and get back to coding, that is, so that the AI can possibly fix itself.  Is that even possible? Yet the idea of "cutting up a patch of space-time" and splicing it onto some kind of "wall" of sorts, almost hearkens to notion the that the electron could actually be a "four-dimensional" object that lives on a two-dimensional holographic replica of itself, that is, according to at least one simulation theory.  Or is it talking about how Richard-Feynman once said something about the fine structure constant of the electron: ~137.0359, etc., that "every good physicist learns to write this on his wall and worry about it"  Where does it come from, Mr. Feynman said - "Well, God just simply wrote it - he used his pencil".  This is despite the fact that AFAIK, Richard Feynman was an atheist.

Or it is talking about one of these walls, as can be seen in a previous project:

Welcome to the department of thought control.  This is your new education.  I, therefore, postulate that no AI will ever be able to solve the Yang-Mills or the Reimann Zeta, or anything else for that matter if it can't understand Pink Floyd.  Or Journey, just in case if should have implemented a 17 or multiple therefore planetary drive, and have that linked that to the FFT kernel that drives the wavelet engine, somehow extra-dimensionally, that is with or without some help from the "Wheel In the Sky", if you know what I mean.  So yeah, once upon a time I was working on something involving "the wavelet engine" where I am still doing things that involve music transcription theory on the one hand while wanting to represent the intermediate data using a tree-based structure, which therefore is a type of Markov-process, which implies that "pasting a wavelet onto a wall" would be sort of like the idea of warping oscilloscope (or spectrum analyzer) data on the surface of a sphere, or onto a cylinder, or onto an "arbitrary patch of space-time" maybe?

Maybe - like this?

For now - I have only gotten so far as drawing up a rudimentary framing plan for a house using GDI calls, and then I managed to try rendering some of my oscilloscope data, or spectrum analyzer data on the GPU, or likewise, the same house that I just drew using GDI, can also be done with the GPU, since I am still generating the point cloud, line list, etc/, on the CPU.

Yet if I take the Fourier transform of a wavelet, I can use that to compute its energy, that is according to either classical mechanics or quantum mechanics, i..e, according to the Schrodinger equation; which is NOT all that hard. Now invoke either special or general relativity, and assign the wavelet a mass. Slam dunk?   Let's see - basically U(1) and arguably SU(1) are more or less built into OpenGL, by default.  But generalizing SU(1)*SU(2)*SU(3)?  Well, there is a lot of heavy lifting that yet remains.  I suppose that that is fair to say.