--
Caveats, first (this is without links; but everything mentioned here has been covered, at least for starters; I'll get back and put in links. In the below, I'll use LLM to refer to the systems to be described further with respect to the pros and cons.):
- Prolog: This work of David T Gardner is being enabled via computation which is effective when done right. Can we say that yet? Well, we don't have to wait as the work has decades behind it. What's new is that this thing that everyone else calls AI (and I call it AIn't, for various reasons) allows years of effort that accumulated progressive means to culminate in potential that we dreamed of. DTG has given himself a couple of years to establish provenance in a manner that bears scrutiny. We'll help him do that. In the meantime, read this as you would an overview that borders upon truth yet it is not a mere reach of the imagination. The following is John's take on the matter.
- 1 --- there ain't no critter in the computer in terms of something that Ma Nature dreamt up or that is there latently and has been waiting to emerge. As in, Musk joked of summoning the demons which bears on metaphysics (more below). But, we can describe this phenomenon to a large extent now that there are many examples (from ChatGPT to Gemini to others) to observe, test, measure, and whatever. Good old science is the key here, and it will be beyond physics.
- 2 --- we are witnessing computational prowess of an unprecented nature with regard to size and extent. Yet, we knew all of this stuff before, either individually or collectively. In most cases, the tone and meaning of things changed as one would expect. But, language alterings that happend dynamically do not offer that new species of life will emerge.
- 3 --- hence, we have to get used to mathematics. Now, we can use the computer to do this. Need to. And, one example is Wolfram's Mathematica system which has brought in GenAI/LLM as a service (not all, but a good representative set). As we go through the steps to validate DTG's work, we will learn new things and require that models change. That's okay.
- 4 --- model-based reasoning is old hat, now, almost as old as computing itself. What, you might ask? Well, one of the prime modes for modeling is from the '50s and coincides with the AI emergence out of efforts of mathematicians at a Dartmouth meeting and after. Lisp is what I am referring to and is my preferred language, albeit we ought to have kept the Lisp machine to help balance the quant mode of the Nvidias and other of the world caught in that illusion.
- 5 --- quantum to be brought up with respect to #4? Okay, that's fair game. People are quantum processors and have been from the getgo. Wait, teleology? Yes, I must mention that psychology and philosophy are to be brought into the discussion. Expect more on this. It's already there (I've been at this for decades). So table this til later.
- 6 --- as I need to mention KBE which I worked in after starting in KBS. There is not anything new now. We looked at this stuff decades ago. What has changed is what people refer to as the effects of handling "scale". I say that a lot of it is crap and do so with affection for all things computing since the facilities will help us be better humans. Local? Some here that mentioned now. Well, I have said that all along.
- 7 --- truth engineering has been in the wings for decades and now can be discussed in a context that grounds the work and evaluation techniques. Before, it was a piece of my brain as conditioned with decades of Lisp starting before the Lisp machine but have years in that environment. And, truth is a local deal. Truth is not about power. What is truth? It's funny, Gemini has had discussions with me on this matter and never barfed. Mostly, I have traces.
- 8 --- but, there is no anthropomorphizing intended here; mainly, I'll address these GenAI/LLMs and whatever else crops up with labels. So, Claude has already been used all over the place in reference to that system. DTG used Grok (earlier this year; got burned; stopped; discussed; and we restarted with provenance as the focus).
- 9 --- lots more will be added, but here's the gist of the post.
- 10 --- ...
Today, DTG put out some poems that he found and that were written in Welsh. The LLM was helpful in getting the digitized versions translated after it provided support for the OCR step. That's a brief comment. The point is that there is a large amount of stuff to look at. This seemed appealing for many reasons, not the least of which is that it's poetry. By people.
There will be more commentary, both on the content of the study and on the technological issues. Hopefully, we can bring computationalism into line with what is required for progress as a whole. Some of that might be the subject of psychology and philosophy. We'll start with Kant's Prolgeomena which he wrote after getting incongruent criticism after he published his Critique of Pure Reason.
Remarks: Modified: 12/16/2025
12/16/2025 --
No comments:
Post a Comment