- From: Owen Ambur <Owen.Ambur@verizon.net>
- Date: Tue, 26 Jan 2021 13:49:30 -0500
- To: Paola Di Maio <paoladimaio10@gmail.com>, Chris Fox <chris@chriscfox.com>
- Cc: W3C AIKR CG <public-aikr@w3.org>
- Message-ID: <29fb6eef-1ccd-053d-3d04-44c8337af908@verizon.net>
Paola, from the project description, it seems like CSAIL is trying to infer goals from behavior but, of course, behavior must be "represented" (documented) in some way that AI agents are capable of interpreting and analyzing. So, yes, it would be good to apply AI to help people understand the goals that are implicit in unstructured (immature) records. One of the interesting points of CSAIL's project description is the possibility that goals may not be perceived as goals by AI agents unless they are achieved. A related issue is that perceived intentions may not really be goals unless there is some way of knowing whether progress is being made and when success has been achieved, i.e., if no performance indicators have been specified and are being tracked. And that points to the primary way in which it seems to me that AI can be helpful in the long run: When public objectives are documented and measured in open, standard, machine-readable format, it should become fairly easy for AI agents to learn and advise human beings what is required to achieve those objectives, thus relieving us from having to learn how to do so over and over again. Then we can focus on how to improve performance rather than trying merely to understand what it is. Moreover, to the degree people may not know precisely what they want to do, AI-enabled services can help them: a) consider the options based upon the personal values that are most important to them and b) connect in partnerships with others in the pursuit of common and complementary objectives best supporting their values. In the meantime, however, the intelligence built into the StratML schema can take us a long way in that direction, through query services (like Chris') that leverage the semantics and structure of the schema. See, for example, the screen shots of a query of the Town of Hilton Head's 404-page comprehensive plan in Chris' app shown in PDF <https://stratml.us/fox/StratNavAppQueryFeature.pdf> and Google Docs <https://docs.google.com/document/d/1XQ7J1EO-NKq_t4_33Bhl4gWlXOGnyDJwotka6FjP4oY/edit?usp=sharing>. His app essentially meets the query requirements but is limited to one document (in the free version) and to one's own collection of documents (in the paid version). Also, it opens the queried elements in edit (form-based) mode rather than in view-only mode in the broader context of the document itself, as per these links to the first <https://connectedcommunity.net/hhi/HHIOP20200304.xml#_4cd6c25c-5f09-11ea-aa27-f0ba1283ea00> and second <https://connectedcommunity.net/hhi/HHIOP20200304.xml#_4cd555c0-5f09-11ea-aa27-f0ba1283ea00> query results shown in the screen shots of the edit/form view in Chris' app. It would be great to see such query/view-only capabilities applied across all >5K of the files in the StratML collection. Owen On 1/26/2021 1:29 AM, Paola Di Maio wrote: > i think you are right Chris > how silly of me- > what about: > using the algo to infer the goal from any non xml doc, then convert to > stratml > would that be useful? > maybe thats what I was thinking > pdm > > On Tue, Jan 26, 2021 at 2:23 PM Chris Fox <chris@chriscfox.com > <mailto:chris@chriscfox.com>> wrote: > > It wouldn't be much of an algorithm if it couldn't infer goals > from plans in which they are clearly demarcated in XML? > > I think what you'd really want to do is run the algorithm against > anything that /wasn't/ StratML and see if it could produce StratML. > > On Tue, 26 Jan 2021 at 02:46, Paola Di Maio > <paoladimaio10@gmail.com <mailto:paoladimaio10@gmail.com>> wrote: > > Maybe Chris can run the algo on a stratml set > > On Tue, Jan 26, 2021 at 10:12 AM Owen Ambur > <Owen.Ambur@verizon.net <mailto:Owen.Ambur@verizon.net>> wrote: > > Yes, Paola, it would be great to see what AI/ML algorithms > might be able to do with the existing StratML collection, > which now comprises >5K files ... but even more so if and > hopefully when public agencies start publishing their > /performance reports/ in open, standard, machine-readable > format... as U.S. federal agencies are ostensibly required > by law to do. > > I'm always on the lookout for partners who might be > willing and able to begin to demonstrate such capabilities. > > While the initial benefit of enabling taxpayers to see > what they are getting for their money will be great, > imagine how AI agents can help agencies learn from failure > and thus improve their performance over time. > > It is painful to watch agency leaders continue failing to > capitalize on that potential. > > Indeed, recent direction from the Trump administration's > OMB director on the way out the door > <https://www.linkedin.com/feed/update/urn:li:ugcPost:6701562085794492416?commentUrn=urn%3Ali%3Acomment%3A%28ugcPost%3A6701562085794492416%2C6757844681394032640%29> > goes so far as to imply that agency leaders have no > accountability for most of the objectives with which they > are entrusted, as if those objectives are merely jokes > being played on taxpayers. Unfortunately, all that seems > to matter is what suits The Politics Industry. The > question is how long voters and taxpayers will put up with > such behavior. Hopefully, not indefinitely. > > Owen > > > On 1/25/2021 7:19 PM, Paola Di Maio wrote: >> Thank you Owen >> wouldn't it be great to try the algorithm on some stratml >> resources >> >> >> On Tue, Jan 26, 2021 at 12:04 AM Owen Ambur >> <Owen.Ambur@verizon.net <mailto:Owen.Ambur@verizon.net>> >> wrote: >> >> "In the quest to capture ... social intelligence in >> machines, >> researchers from MIT’s Computer Science and >> Artificial Intelligence >> Laboratory (CSAIL) and the Department of Brain and >> Cognitive Sciences >> created an algorithm capable of inferring goals and >> plans, even when >> those plans might fail." >> >> "... ability to account for mistakes could be crucial >> for building >> machines that robustly infer and act in our interests >> ... Otherwise, AI >> systems might wrongly infer that, since we failed to >> achieve our >> higher-order goals, those goals weren’t desired after >> all. We’ve seen >> what happens when algorithms feed on our reflexive >> and unplanned usage >> of social media, leading us down paths of dependency >> and polarization. >> Ideally, the algorithms of the future will recognize >> our mistakes, bad >> habits, and irrationalities and help us avoid, rather >> than reinforce, them." >> >> https://scitechdaily.com/new-mit-social-intelligence-algorithm-helps-build-machines-that-better-understand-human-goals/ >> <https://scitechdaily.com/new-mit-social-intelligence-algorithm-helps-build-machines-that-better-understand-human-goals/> >> >> Wouldn't it be nice if AI-assisted business >> networking services helped >> us avoid polarization and needless dependencies on >> The Politics Industry >> as we strive to achieve public objectives documented >> in an open, >> standard, machine-readable format? >> >> https://www.linkedin.com/pulse/politics-industry-v-we-people-magic-formula-owen-ambur/ >> <https://www.linkedin.com/pulse/politics-industry-v-we-people-magic-formula-owen-ambur/> >> >> Owen >> >> >> > > > -- > Chris Fox > Chris C Fox Consulting Limited > chris@chriscfox.com <mailto:chris@chriscfox.com> > +44 77 860 21712 > <http://www.chriscfox.com> > <https://calendar.x.ai/chriscfox/freeconsult> > <https://www.linkedin.com/in/chriscfox/> > <https://twitter.com/chriscfox> > <https://www.facebook.com/StrategicCoffee> > Have you tried https://www.StratNavApp.com > <https://www.stratnavapp.com/>, the online collaborative tool for > strategy development and execution? > > Chris C Fox Consulting Limited is registered in England and Wales > as a Private Limited Company: Company Number 6939359. Registered > Office: Unit 4 Vista Place, Coy Pond Business Park, Ingworth Road, > Poole BH12 1JY >
Received on Tuesday, 26 January 2021 18:49:49 UTC