Misconceptions about what knowledge representation truly is

Over the last couple of days it has become abundantly clear that the
hijacking of the term "knowledge representation" by computer scientists to
mean "a framework for representing human knowledge in a computer, enabling
the efficient execution of intellectual activities such as pattern
understanding and inference" has created some serious problems.

I have indicated in past interventions on K3D with Daniel Ramos that
knowledge, and specifically its machine readable representation cannot
mathematically represent all human knowledge and worse, not even the
functioning of the human brain, and ironically K3D in ensemble with AI
assistants has confirmed this.

To mathematicians like myself and also doubling as a philosopher I have
serious misgivings about this narrow scope definition of what constitutes
the representation of knowledge.

It is exactly this misconception that has the generative LLM tribe running
in overdrive and hurtling humanity into an economic precipice and onto an
environmental cataclysmic path.

I think that a fundamental task of this CG should be to narrow down what
constitutes "knowledge that is mathematically representable in machine
readable format".

Why is this so important? Because the hard problem of consciousness as
pondered upon by neuroscientists, cognitive scientists,  computer
scientists and philosophers hinges exactly on this. Consciousness allows us
to circumvent the mathematical limitations of representing reality by
thinking and expressing in natural language things that cannot be pinned
down mathematically.

That's why myself as a Godelian school mathematician resort to philosophy
and neuroscience to try to figure out how we can best formulated knowledge.
And in this process, borne out by K3D and its ensemble of AI assistants,
only domains of discourse will do the trick, not tokenization.

I feel Daniel is on to something with K3D and he has now compelled me into
deep diving into semiconductor and microprocessor design and the Nvidia
GPUs and the CUDA platform, for which I downloaded literature several
months back.

I am glad Daniel is an electrical engineer and myself a mathematician,
because maybe we can steer most of the Gen LLM tribe of computer scientists
back on track.

Knowledge representation as a "domain of discourse" isn't "owned" by
computer scientists, and it is high time the record is set straight.


Milton Ponson
Rainbow Warriors Core Foundation
CIAMSD Institute-ICT4D Program
+2977459312
PO Box 1154, Oranjestad
Aruba, Dutch Caribbean