Re: reminding the scope for this AI KR CG

Yes, Paola,  I agree with you and Dave that natural language in the end is
the primary focus for knowledge representation.

But here is the kicker. The Big Boys in Silicon Valley or Big AI Tech are
focused primarily on getting their versions created with all of humanity
subservient to a White, Male, Christian and English speaking dominated
world of science and data science and even engineering.

I deal with ICT4D and AI for Good issues for my main focus groups,
indigenous peoples and Small Island Developing States.

Data sovereignty and  (cultural) knowledge sovereignty are very important.

I do not have to stress that the total of all human knowledge is captured
in written and orally transmitted form natural language.

The article in Aeon nails it perfectly.

https://aeon.co/essays/generative-ai-has-access-to-a-small-slice-of-human-knowledge
.

If we ignore this we are wasting our time.

It is precisely this loss of knowledge which resides mostly in
"specific/restricted/limited domains of discourse" that will make AI not
the "beneficial technology for all of humankind " but the destroyer of
plurality and pluriformity of human knowledge.

My mandala graph framework is an intent to show exactly where the
limitations of formal systems lie and how we can proceed to capture all
human knowledge available in natural language.

Daniel Ramos, myself and I assume quite a few other members of this AIKR CG
agree that KR for AI cannot be developed for its own sake, but with a
specific purpose in mind.

That purpose is and should be for all natural language in all languages, in
written, oral and other formats.

I have networked with native American AI data and knowledge sovereignty
initiatives,  the IFLA, SIL (www.sil.org) and SIDS on indigenous and creole
languages.

Generative AI based on LLMs is creating a world of linguistic impoverished
knowledge and even creating a DoubleSpeak.

Without retaining the linguistic pluriformity and plurality creativity and
the rich diversity in cultures will cease to exist and we will be doomed as
mankind.

Social scientists,  psychologists, philosophers and linguists all have
pointed out the imminent danger of what AI is creating in this respect.

So to me the purpose of KR for AI should stress the "creation of formal or
rational expression of all knowledge in natural languages in written, oral
or other form, captured in machine readable format to allow artificial
intelligence driven computational processing for specific applications or
tasks in specific domains of discourse".

KR for AI should have maximum "semantic compression ". The current course
of GenAI is exactly towards the opposite.


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

On Tue, Nov 18, 2025, 07:39 Paola Di Maio <paola.dimaio@gmail.com> wrote:

> When I launched this CG, several years ago, the horizon was mixed. I was
> open to explore what could lie ahead, looked at how KR would help
> detangle  the new wave of ML AI.
> To do that, we needed to be able to put our finger on what exactly KR was.
> And that was tough. There was no single resource
>
> Now we have a bunch of vocabs and conceptual spaces *albeit still in draft
> version that define with more precision
> the KR knowledge domain, and that is helping us to navigate and stay on
> top of the rapid advances in AI
>
> Vocabularies are units of computation in language models!  Heck yes
>
> Natural language processing is based on natural language terms
> A first report was published summarizing the critical path this CG is
> following to achieve that goal and then we brew bits here and there
>
> I can understand that newcomers, or anyone who has not been paying
> attention as much as others as to what this CG is actually doing and why,
> or are not interested  in it for whatever reason may struggle to grasp with
> what the CG is all about and may want to suggest other directions,
>
> The scope of work has now become more focused on vocabs and concept maps
> that define KR as a primary domain of interest for reliable AI
> The next step
> is to attempt to publish these as schemas/specifications.  We are not
> planning beyond that at this time.
>
> Natural language is the thing here.
>
> Domain specific knowledge representation is not in scope, although there
> is now a conceptual space for domain KR *the bubble on the left
> pointing clearly to Domain ontologies, possibly in scope of another CG?
> and there may well be terms in our core vocab pointing to domain specific
> knowledge. Room for analysis there!
>
> I may not be able to read/reply to emails that raise issues not in scope,
>
> This is crunch time for this CG, which needs to keep its eyes on the
> natural language for KR ball
>
>  *it goes without saying that pertinent contributions including
> intelligent responses and constructive critiques are welcome and will be
> included in the report. Please get in touch  if you wish to make
> contributions to be considered for the report or for further refinement
>
> Participants working on other things, please consider proposing a new CG
> that follows your vision
>
> Please refer to the updated diagram below
>
> Second report with updates from TPAC will be out before the end of the
> year, I hope, with all updates
>
>
> Thanks for your continued interest and participation, exciting things are
> coming to us all!
> [image: AI KR VOCABS NOV 2025.jpg]
>
> PDM
>
>
>