Re: definitions, problem spaces, methods

Hi Dave,

I've enjoyed our interaction, and will call it quits with this response. 
See below:

On 11/9/2022 4:15 AM, Dave Raggett wrote:
> Hi Mike,
>
> GPT3’s weaknesses in respect to coherence, etc. is unsurprising and 
> should encourage research on moving on from text and image prediction 
> to work on reasoning and continuous learning. This in turn motivates 
> work on imprecise and imperfect knowledge, where I expect neural 
> networks will shine. That includes the role of causal explanations as 
> a basis for learning, and the ability to reason about past, present 
> and imagined situations, including the beliefs of others. I very much 
> believe that hand authoring of knowledge representations will give way 
> to machine generated KR.

I, too, hope we can move away from hand authoring of KR and progress to 
machine-generated KR. My operating belief, however, is that such machine 
generation will not start de novo, but relies on well-vetted and 
well-reasoned reference knowledge representations, in part in the form 
of knowledge graphs, to bootstrap the efforts. We also need more 
attention to reasoners and a suitable knowledge representation. GPT or 
similar meets neither of those tests.

For machine reasoners and machine logicians to work, I believe we will 
need reference starting representations from which to train these 
learners. Consistent with that belief, I do not think that generative 
models that begin from unsupervised bases are the correct way to 
bootstrap this process. This is the motivation behind my own work and 
contributions in KBpedia <https://kbpedia.org/>, though I do not claim 
it is yet ripe for such purposes from a reasoning and logic standpoint. 
Hence, my ongoing interest in Charles Peirce. ;)

> I hope you can respect my view that KR is a subset of AI, as in the 
> Wikipedia definition.

I certainly do respect your opinion, one which I used to hold and which, 
I would guess, 80-90% of current semweb and KR practitioners agree. As I 
said in the conclusion to my book [1]:

    "When I began this book, I blithely assumed that knowledge
    representation was a subfield of artificial intelligence. Every
    taxonomy that I have seen about AI subfields and that included
    consideration of knowledge representation shows KR as a subsidiary
    field. I frankly had never questioned the relationship."

    "However, when considered, mainly using prescission [a very powerful
    logical operation proponed by Peirce], it becomes clear that KR can
    exist without artificial intelligence, but AI requires knowledge
    representation."

Perhaps some day you may have a similar epiphany, but if you or others 
don't, that is not cause to withhold respect. ;)

>  The media hype around AI is frustrating in that attention grabbing 
> headlines detract from a clear exposition of the underlying concepts.
>
> p.s. you may be interested in the talk I’ve prepared for next week’s 
> NSF/EU workshop on research priorities.
>
> https://www.w3.org/2022/11/Raggett-AI-Priorities.pdf

Thanks for sharing. This is not an approach or set of luminaries I would 
follow. I think there is much to be said about the free energy 
approaches of Karl Friston (which I also believe can be related to many 
Peircean ideas). I hope to be able to say more about this in the coming 
months.

Good luck to your endeavors!

Best, Mike

[1] https://www.mkbergman.com/a-knowledge-representation-practionary/