AI for Planning & Scheduling

Among AAAI’s topics, the one on Planning and Scheduling <https://aitopics.org/search?filters=taxnodes:Technology%7CInformation%20Technology%7CArtificial%20Intelligence%7CRepresentation%20%26%20Reasoning%7CPlanning%20%26%20Scheduling>  seems especially ripe for progress.  See especially these parts of the description:

 

One approach is to generate expectations: facts that should be true during different stages of a plan's execution. 

 

In planning parlance, such “expectations” are called milestones or performance indicators of the target results type, on a timeline.

 

When an expectation is violated, a discrepancy occurs between the expected and perceived facts. 

 

That is, the actual, measured result differs from the target result.

 

A new trend in autonomy is to include goal reasoning capabilities. In the event of a failure, the original goal may no longer be warranted. Perhaps robust autonomous agents need to generate and change their goals in response to a changing environment.

 

Demonstrating capability to achieve goals is the very essence of “intelligence” – including recognition not only when the original goals may have been unrealistic but also when they are affected by changes in circumstances.

 

The first requirement is to document plans in both human- as well as machine-readable format.

 

https://www.dictionary.com/browse/the-best-laid-plans-of-mice-and-men-often-go-awry 

 

We human beings can use all the help we can get not only to clarify our goals but also to modify them when our assumptions and/or perceptions prove faulty (and we don’t want to admit it to ourselves, much less to others).  While we are all prone to honoring sunk costs, our public agencies and the political processes are particularly ripe targets for performance improvement. https://www.psychologytoday.com/us/blog/anxiety-files/201409/letting-go-sunk-costs 

 

Presumably, AI agents might be programmed free counterproductive, face-saving biases.

 

Owen

 

 

From: Vinay K. Chaudhri <vinay_chaudhri@yahoo.com> 
Sent: Wednesday, October 31, 2018 8:51 AM
To: public-aikr@w3.org
Subject: Re: Fwd: Some comments on the description of this group

 

Paola,

For R&N, please see: https://www.amazon.com/Artificial-Intelligence-Approach-Stuart-Russell/dp/9332543518/

For AAAI's efforts in this area, please see: https://aitopics.org/misc/about

There is, obviously, room for contribution.

These days to impact web standards, a viable approach seems to be to join an internet company (or at least, start by securing their support :-)

Best,

Vinay.

 

On 10/31/2018 12:25 AM, Paola Di Maio wrote:

Vinay

 

Thanks a lot for your questions, see my comments below  in italics

but of course , everyone, please feel free to address Vinay's qs as you see fit

although I initiated this, it belongs to all members

 

~~~~~
The overall goal/mission of this community group is to explore the requirements, best practices and implementation options for the conceptualization and specification of domain knowledge in AI.
~~~~~
"Conceptualization and specification of domain knowledge" reminds of the expert systems era and a family of techniques that have little overlap with the current ML approaches. While in silicon valley, ML and AI are largely used interchangeably, those with some historical perspective, recognize the evolution of techniques. For consider this video that explains the DARPA perspective on AI to encompass three phases of AI. https://www.youtube.com/watch?v=-O01G3tSYpU 

 

thanks 


It will be helpful for this group to refine its description of the set of AI techniques that are of interest.

 

thanks, I ll watch and take input but everyone feel free!


~~~~~~~~~~~~~
We plan to place particular emphasis on the identification and the representation of AI facets and various aspects (technology, legislation, ethics etc) with the purpose to facilitate knowledge exchange and reuse.
~~~~~~~~~~~~~
The above statement, even though, appears agnostic to the specific technique being used, but really pertains to the second phase (ie, ML-oriented) techniques.

 

uhm, why do you think so? If anything  it refers to multifaceting of  knowledge domain modelling techniques (thats where I get the idea from) but it could be that multifaceting also used in ML? 

 

  These days we do not hear too many people talking about the legislation and ethics around the first-phase (ie, rule-based) techniques.

part of the problem we are trying to address, perhaps? 

~~~~~~~~
Therefore the proposed outcomes could be instrumental to research and advancement of science and inquiry, as well as to increase the level of public awareness in general to enable learning and participation.
~~~~~~~~~
There are other efforts to educate public about AI. E.g., there is an AI for K-12 initiative, and AAAI engages in considerable outreach to general community. It is not clear to me why a W3C group needs to be devoted to this problem.

the way I see it:  

'education' is not just for schools, but for everyone who continues to learn-and we must

* AI (in various forms) are powering our information systems AND the web (the way the web works, and the doesnt work etc)
* W3C is among other things a standardization body for the web
* plus:  W3C community groups and workgroups are  highly collaborative, open to read and write, and fairly inclusive by design
* If  we manage to construct somethingk, this could become a WG and possibly a starndard


These  reasons for me make a unique and compelling argument in support of an open collaborative forum which could serve XAI among other things

do we need a web stardard for XAI?  I think so, maybe

 

~~~~~~~~~~~~~~~~~~~~~~~
Proposed outcomes: 

* A comprehensive list of open access resources <https://www.w3.org/community/aikr/welcome/ai-kr-task-list/knowledge-sources-for-ai-kr/>  in both AI and KR (useful to teaching and research)
* A set of metadata <https://www.w3.org/community/aikr/welcome/ai-kr-task-list/metadata/>  derived from these resources
* A concept map <https://www.w3.org/community/aikr/concept-maps/>  of the domain
* A natural language vocabulary to represent various aspects of AI
* One or more encoding/implementations/ machine language version of the vocabulary, such as ChatBot Natural Language Understanding & Natural Language Generation
* Methods for KR management, especially Natural Language Learning / Semantic Memory

~~~~~~~~~~~~~~~~~~~~~~~~

These sound like useful set of outcomes, but it is not clear who are they meant for, and who is it going to use them?
There may already exist such resources in some sub-areas, e.g., NLP tools, ontology editors, etc.

 

The problem that I think we face is that there is 'no shared conceptualization' for AI (using an old metaphore) and no common language (the same argument that we would make if we wanted to come up with, say an ontology for AI.  I have not seen any resources or public effort devoted to this, but if you do please point is, because a collaboration would definitely be on, I think

 

There is also some degree of coverage of these topics in the standard textbooks such as R&N.

url plz?  the degree of coverage I have seen so far is not fit for purpose (see above) but again, please point us

 

Are you proposing to write a book? build a catalogue?

either/both/neither

I am not proposing anything at the moment, but the ideas above sound good if useful

we could also do a movie :-) (or rather an educational video)

what I personally aim for is to clarify what is AI (for the purpose of accountability and reliability) since there is a lot of hype and confusion being propagated at the moment

 

Should you be working with AAAI on these?

I am a member o AAAI (or I was at some point) is there a discussion group on AI?

A web page? please feel free to enter related efforts in the google form linked on the group home page, or post here 

 

How would you maintain them as things change?

If there are people devoted to the maintenance, future generations can maintain

if nobody is around to follow up, the outcomes would become frozen in time, or fossilized

as a testimony of our restlessness around the issues at hand

 

Tell us more about you

 

 

Best Wishes,
Vinay.