A formalized approach to consider agents' norms and values

[Gary Berg-Cross]

Ken Forbus posted this elsewhere but it should be of interest to this community:

"How can an AI system build up and maintain an accurate mental model of people's norms, in order to avoid social friction? This is difficult because norms not only vary between groups but also evolve over time. Taylor Olson's approach is to develop a formal defeasible deontic calculus, building on his prior work on representing social and moral norms, which enables resolving norm conflicts in reasonable ways. This paper appeared at the Advances in Cognitive Systems conference in Palermo last month."

https://arxiv.org/abs/2407.04869?

Gary Berg-Cross 

Potomac, MD

240-426-0770

[Alex Shkotin]

Gary, thank you!

I have sent it to my favorite formal philosophers https://www.facebook.com/share/SaGkSXTmVF2HcJp9/

Alex

вт, 16 июл. 2024 г. в 18:12, Gary Berg-Cross <gberg...@gmail.com>:

--
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 on the web visit https://groups.google.com/d/msgid/ontolog-forum/CAMhe4f1faWoWBD0iKqvM2%2BL1xK7KfdyXFy4YfDcqk-BO41OJaw%40mail.gmail.com.

[John F Sowa]

Alex and Gary,

The article cited in your notes is based on logical methods that go far beyond anything that is being done with LLMs and their applications to generative AI.  Such developments are important for evaluating and testing the accuracy of the output of OpenGPT and related systems.  

I won't say anything in detail about the cited article.  But I noticed that some of the researchers mentioned in the article participated in the IKRIS project, which was developing metalanguage extensions to Common Logic.   For a summary of that project with multiple references for further study, see https://jfsowa.com/ikl . 

The documents cited there include a list of the participants and the topics they were working on.  The issues they discuss are vitally important for testing and evaluating the results generated by LLMs.  Without such evaluation, the output generated by LLMs cannot be trusted for any critical applications.

John

 

---

From: "Alex Shkotin" <alex.s...@gmail.com>
Subject: Re: [ontolog-forum] A formalized approach to consider agents' norms and values

Gary, thank you!

I have sent it to my favorite formal philosophers https://www.facebook.com/share/SaGkSXTmVF2HcJp9/

Alex

вт, 16 июл. 2024 г. в 18:12, Gary Berg-Cross <:" style="box-sizing: border-box; color: rgb(0, 102, 147); text-decoration: underline; user-select: auto;">gberg...@gmail.com>:

[Polovina, Simon (BTE)]

Hi all.

John, thank you for interrelating the multiple semantics and ontology projects in https://jfsowa.com/ikl, a handy resource!

Regarding the article forwarded by Alex, Stamper’s Semiotic ladder would be pertinent:

http://assets.cs.ncl.ac.uk/seminars/101.pdf (and multiple subsequent materials on the subject: https://scholar.google.com/scholar?q=stampers+semiotic+ladder)

I’ve referred to it in my past work, e.g.:

• A Transaction-oriented architecture for enterprise systems - Sheffield Hallam University Research Archive (shu.ac.uk)
• Semiotic models of trust and usability for agent-managed Grid services - Sheffield Hallam University Research Archive (shu.ac.uk)

I believe that the late Stamper’s work is underrepresented, but it would be interesting to hear your thoughts.

Thanks!

Simon

 

Dr Simon Polovina

Department of Computing, Sheffield Hallam University, UK

Ontolog Forum Profile

[Alex Shkotin]

Hi Simon,

Nice idea in 101.pdf "...treating the notion of a sign as a primitive concept, we can arrive at much clearer definitions of many of the troublesome concepts in the information systems domain, not least the concept of information, itself."

The question is what other primitive concepts and relations we need to axiomatise Information theory?

Keeping in mind as a benchmark for example Hilbert's axiomatisation of Euclid's Geometry.

The topic raised in the article brought by Gary: 

"...to maintain an accurate mental model of other's norms, and thus to avoid social friction, such conflicts must be detected and resolved quickly. Formalizing this process has been the focus of various deontic logics and normative multi-agent systems. We aim to bridge the gap between these two fields here." 

is very specific. This is why I am transferring this article to people who are working in both of these fields "various deontic logics and normative multi-agent systems." They use an axiomatic theory approach directly and thoroughly.

For me an interesting situation may be if we have for the same application area from enterprise to science an axiomatic theory and a formal ontology.

This topic may be close to your https://sysaffairs.org/move/move24-cfp MOVEment ⚖️.

Alex

чт, 18 июл. 2024 г. в 10:00, 'Polovina, Simon (BTE)' via ontolog-forum <ontolo...@googlegroups.com>:

--

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 on the web visit https://groups.google.com/d/msgid/ontolog-forum/DB8PR03MB549759BED56A3513F21AAB97D9AC2%40DB8PR03MB5497.eurprd03.prod.outlook.com.

[John F Sowa]

Simon,

I'm sorry to hear that Stamper had died.  I had regarded his work over the years as quite respectable. 

Most readers are unlikely to click on titles with URLs.   But if you include an interesting quotation (one or two paragraphs), they are more likely to click.  Perhaps you could elaborate on your note below.

John

 

---

From: "Polovina, Simon (BTE)' via ontolog-forum" <ontolo...@googlegroups.com>

Hi all.

John, thank you for interrelating the multiple semantics and ontology projects in https://jfsowa.com/ikl, a handy resource!

Regarding the article forwarded by Alex, Stamper’s Semiotic ladder would be pertinent:tp://assets.cs.ncl.ac.uk/seminars/101.pdf (and multiple subsequent materials on the subject: https://scholar.google.com/scholar?q=stampers+semiotic+ladder)

[John F Sowa]

Auke,

My formal education emphasized math & logic, but I designed conceptual graphs as a logical foundation for mapping computational data to and from natural languages.  Languages have always been one of my strong interests.

All four of my grandparents migrated from Poland around 1900 to 1905, and my maternal grandmother lived with us.  So Polish was the language I heard and spoke all day long while my father was working.  When you start with two languages, learning a few more is easy.

Peirce's father taught him mathematics, Greek, and Latin from early childhood.   I developed my first version of conceptual graphs while I was taking Minsky's course on AI and a course on psycholinguistics.  When I came across Don Roberts' book on existential graphs, I saw the parallels between CGs and Peirce's EGs.  Since Peirce had also addressed the mappings between language and logic, I adopted similar methods.   For the problems and challenges, see the concluding chapter 7 of my book on conceptual structures:  https://jfsowa.com/pubs/cs7.pdf .

I met Stamper at various conferences, and I found his outlook and methods to be congenial.  He addressed the mappings between natural languages and computational representations in a way that could be mapped to and from EGs or CGs in similar ways.  I didn't map CGs to his charts directly, but his methods address the issues of choosing rigid formal patterns (graphs or charts) that can approximate flexible informal words and phrases.

And by the way, these issues are the primary reason why I am very critical of the claims about LLMs as a foundation for intelligence or cognition.   Most of the people who develop them are unaware of (or deliberately ignore) the challenging problems of mapping any kind of formalism to and from natural languages.

John

 

---

From: "Auke van Breemen" <a.br...@upcmail.nl>

John,

 

It is a pity indeed that he passed away, he was a scientist of the constructive type.

Since I always saw a similarity in your and Ronalds approach in how you deal with Peirce maybe it is a good time ta ask your opinion about this.

 

Both of you differ from the main sentiment on peirce-list, which concentrates on interpretations on what peirce might have ment. So the theoretical fight is about the right interpretation of the masters writings. Concentration on the signs in itself, seems an adequate label.

 

Alternatively, the both of you use Peirce in order to contribute to his goal with semiotics, which is to understand interpretation processes as sign processes. But, the perspectives differ offering different views on the processes.

 

In my opinion you concentrate on the semantic dimension, while Ronald had a keen eye for the pragmatic dimension, the pragmmatic and social layers of his ladder. This shows in his ontology charts in which always society is the most remote root node. Thus it becomes possible to adress the problem of differing knowledge systems in any actual situation, hence also Ronalds heavy interest in 'responsible behavior'.  

 

I am curious after your view on the relation between your graphs and Stampers charts.

 

Best,

 

Auke van Breemen

 

PS. In https://link.springer.com/chapter/10.1007/978-3-319-94541-5_10 I elaborate on Stampers take.

[Polovina, Simon (BTE)]

Thanks, Auke and John, for your inputs, which are enlightening.

In my A Transaction-oriented architecture for enterprise systems - Sheffield Hallam University Research Archive (shu.ac.uk) paper, I refer to Stamper’s Semiotic Ladder thus (pp.73-75):

“…the ‘cost imposing’ layers in [Stamper’s] ladder structure illustrate the areas in which the productivity of computers (‘the technology platform’) benefits information (enterprise) systems much better than manual systems. It is where computers are much better than humans, hence the success of technologies such as data processing that we cannot imagine being without nowadays. The ‘value producing’ layers are where humans are better than computers. Classical experiences with Artificial Intelligence have shown how poorly computers perform…” [note: I wrote the article in 2013.]

If you visit those pages through the above link, you’ll notice I used Conceptual Graphs and Formal Concept Analysis to convey the broader dimensions of balancing business transactions, e.g. Psychological Stimulation (which, as a side note, I incorrectly—or perhaps correctly—refer to as Psychical Stimulation 🤔 ). These broader dimensions arguably resonate with Stamper’s ladder.

Cheers,

Simon

 

Dr Simon Polovina

Department of Computing, Sheffield Hallam University, UK

Profile │ Email

 

From: Auke van Breemen via CG <c...@lists.iccs-conference.org>
Sent: Monday, July 22, 2024 11:02 AM
To: CG <c...@lists.iccs-conference.org>
Subject: [CG] Re: Stamper's semiotic ladder

 

John, list,

 

Thanks, this answer offers the possibility to address Simons ladder question. 

 

In my opinion the semiotic ladder didn't get the attention it deserves. 

 

Also for the reason John wrote about:

And by the way, these issues are the primary reason why I am very critical of the claims about LLMs as a foundation for intelligence or cognition. Most of the people who develop them are unaware of (or deliberately ignore) the challenging problems of mapping any kind of formalism to and from natural languages.

--

 

The difference in the ladder between the IT layers (exclusively technical, see Hall) and the higher layers (beside the technical we find formal and informal elements, following Hall again) address this problem. The distinction of the antropologist Hall between technical,formal and informal culture is pointed at by Stamper. This concerns the translation problem.

 

But not only for this reason. A second challenge for LLMs is to deal with differences in goal orientation. An identical question may be asked for different goals and also the proximate goal of a question is an answer, but the remote goal remains hidden unless explicated. The social and pragmatic layers of the ladder provide a scaffolding for dealing with those issues. 

 

Interestingly Stamper started with responding to Morris threefold distinction between a syntactical, a semantical and a pragmatical level. Due to his work in IS, he felt obliged to enlarge and refine the list. This resulted in his ladder. 

 

The group around Kecheng Liu in Reading continued Ronalds work, working their way further up the ladder. Eventually resulting in work on the Assessment of pragmatic interoperability for process alignment. But still this is work done from the perspective of an IT system as an affordance and it only covers horizontal alignment (between similar layers).

 

In https://link.springer.com/chapter/10.1007/978-3-319-94541-5_10 (The social layer of Stamper ladder, just scoll down after the link to get to the text.) I axplore the social layer.

 

Abstract

The social layer of Stampers ladder addresses the problem of responsible agents interacting with each other. It is the layer at which in organizations decisions are made and transformations negotiated. The method we present supports this human interplay. It combines principles of actualism, ontology charts, the knowledge in Formation process model and the Cynefin framework to gather and combine quantitative data with qualitative data, expressing attitudes and perceptions in meaningful diagrammatic representations of business processes. The analytic tool Sensemaker can be used to support decision making.

--

 

At the same time the ladder needs an extension if we want to shift focus drom IS as a system/object, to an interaction of IS systems/ a semiotic definition of information processes. For that we need Peirce's theory of interpretants as I argue in https://pdfs.semanticscholar.org/6e1e/b1d74909a117c71d89e8c697c382c7ab7137.pdf.

 

 

In daily use the ladder offers a possibility to ask for vertical alignment. This implies that we must be specific about what can be rigidly defined and what not. The ontology chart mediates between my vague approach in the social layer and Stampers S(emantic)AM (the pragmatic /semantic layer , from this we can go to CG. For one part in order to go down the road towards technial realisation. but also to loop back to the social layer. For which we again need Stampers work.

 

Best,

 

Auke van Breemen 

 

PS LLMs cannot deal with the dicent or also dynamical interpretant position in the KiF scheme; it is the moment in the interpretation process the specific goals and context/history of the interpreting system comes ino play in getting at a conclusion or response. see p 93 in https://www.scitepress.org/papers/2010/32587/32587.pdf  

 

Op 21-07-2024 03:38 CEST schreef John F Sowa <so...@bestweb.net>: