Summary and next
X DdC
Summary and next
Summary of 'Ontology families for AI-systems'
https://www.ontooo.com/Ontology.pdf
Joining the Ontology-Forum should have been done earlier. My
advertised paper yielded approvals. John Sowa's 'Chat-systems have no clue
They cannot apply the AI crown-jewels from the preceding 60 years,
they cannot answer questions about novel mini stories, they are
autistic, have no moral compass, no feelings, no emotions, no
existence in space with our embodiment, no awareness of the female -
male differences, etc.
Due to lack of morality they regurgitate the white lies from the media
of the preceding century - with unacceptable consequences -- also ignored.
[[ Responses on the 1972 'Limits to Growth' world simulator: There are
no limits/ The market will deal with it/ Human ingenuity will deal
with it.
No: Sapiens/the media have excelled in Pathological Optimism &
Suicidal Generosity. ]]
<<<
A chat-system produced a 49 line reply to the prompt:
How to justify the $40K public education budget child/year in
NYC?
that ended with:
In short, there is broad agreement on where the money goes; there
is much less agreement on whether the outcomes justify the cost.
Lacking morality should have produced the reply that a response is not
feasible.
Missing in the response is a comparison against budgets elsewhere: in
California, for example, the budget is around $19K child/year. This
would have provided a critical commentary instead of the hedging,
white-wash reply, with its bias flavor - and thus should have been
avoided.
>>>
The chat-systems pretend they can provide physical medical
diagnosis and mental support with unacceptable consequences; see the paper.
The good news is the emergence of neuro-symbolic systems by
companies (Palentir, DataBrics, Permion, etc.) that have or create
small ontologies and may use Mini-Language-Models to populate the
ABox-es of these ontologies.
I want more:
We need neuro-symbolic systems driven by an LLM by using a multitude
of ontologies: one for 'dummies' with the additional ability to
delegate to the numerous others suggested by Steve Pinker in 'The Blank
Slate/ The modern denial of human nature'.
Knowledge/ information is for individuals already large and diverse.
This size suggests the maxim 'divide & conquer'. Pinker suggests to
recognize the following decomposition based on domains [Pinker]:
- intuitive physics
- intuitive biology/ natural history
- intuitive engineering
- intuitive psychology
- spatial sense
- number sense
- probability sense
- intuitive economics
This decomposition can be applied - by analogy - to an ontology for
AI-systems. Hence it is advised to create micro-ontologies for the
different domains in chat-systems similarly. Since these domains can
still be large further decompositions in sub-domains is recommended -
consider, for example, the different specialization in physics, biology,
engineering, etc.
This requires indeed a module that can pre-process input to decide
which micro-ontology to engage. To be done by a foundation ontology?
Meta ontologies (plural indeed) must be part of the mix as well.
This assumes that we have developed, incrementally, the TBox-es of
these ontologies (with concepts) for selected domains. Subsequently
the generation of an LLM should generate the contents for their
ABox-es as side-effects.
Hence I propose that we (= the Ontology Forum members - and others
recruited) create those TBox-es. This is like the effort of Doug
Lenat, but this time with the recognition that AI (after 60 years)
MUST have ontologies to support all future AI-systems.
How to achieve that goal?
Alphabet/ Google has created an amazing amount of curated mini answers
(beyond providing relevant links as done in the previous decades).
Somehow we need a task-force/ organization (of which they are a member)
that can leverage these answers (for the generation of Mini Language
Models for incrementally committed domains) for bootstrapping and
subsequently take on the larger challenges.
Does this proposal make sense?
If not, why not?
If yes, what pragmatics to employ for creating a mini-organization
(with who?) that takes the lead?
[ The long time perspective: John Horgan's 'End of Science' suggests
that all low hanging fruits have been taken.
A mega ontology may be required to generate the next breakthroughs. ]
dennis de champeaux
Silicon Valley
[[ Surviving my 5th cancer.
My med ontology helped to self-diagnose.
Thankful. ]]
Alex Shkotin
Hi Dennis,
The current state of theoretical knowledge is that it's become very plentiful and quite tightly and intricately intertwined. On the other hand, almost all of it, in one form or another, exists in computers. This means it can be formalized.
Calling these formal theories ontologies is a matter of taste. See https://www.linkedin.com/pulse/ontologyit-alex-shkotin-owg3f.
The most open project for accumulating formalized theoretical knowledge is the Wolfram Project. All the others are closed to a greater (Permion, Palantir) or lesser extent (Cyc, AlphaGO).
How should this vast array of knowledge be created, stored, and expanded?
My humble opinion would be a system of theory frameworks. How many will there be? I think about a hundred thousand.
An example of a framework is here (PDF) Theory framework - knowledge hub message #1. The idea is as follows:
Storing the theory of a particular subject area in one place and maintaining it (including formalization) through collective efforts is easily possible with the modern development of technology. The concentration and verification of knowledge achieved in this case should give a powerful ordering of theoretical knowledge, which will facilitate their formalization, i.e. mathematical notation, and therefore algorithmic processing in many cases, up to the semi-automatic proof of various kinds of consequences, for example, theorems. This message describes what the framework of the theory is, intended for unified storage and collective accumulation of its results.
One form of organizing projects of this class in an open format could be the nascent HG project Holon Graph | Community Groups | Discover W3C groups | W3C.
Good luck,
Alex
--
All contributions to this forum are covered by an open-source license.
For information about the wiki, the license, and how to subscribe or
unsubscribe to the forum, see http://ontologforum.org/info
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-foru...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ontolog-forum/CAJ%2B4qGcZ%3D1h3r3Kp_b%3DQ5_TOO2fSifYfbL39svOELD5JzqWSFw%40mail.gmail.com.