Geothink&Learn 3: Governing Artificial Intelligence

Image result for artificial intelligence

Image courtesy of www.digitaltrends.ca

Cities worldwide are seeing a rapid expansion in the promotion of artificial intelligence (AI) and algorithms to assist in governance. While AI is often heralded as a means of optimizing and making more efficient various domains of city-citizen interactions (such as transportation, real-time citizen notifications, predicting urban change), many of its promises and threats have been hyperbolized in popular media. This panel brought together leading academic experts to discuss the opportunities, challenges, and implications of this transition as they apply to cities and citizens, with a focus on governance and policy in a Canadian context.

What are the emerging AI shifts to city planning and design? How is AI shaping governments and governance? What are the current factors shaping AI policy? How do we hold AI accountable, especially when we don’t know how AI works? What is the impact of public-private partnerships that emerge out of AI and smart city adoption? What are the social implications of AI, and how can we better regulate to prevent bias? 

On Thursday, December 7 at 12:00 (EST), Geothink.ca held its third monthly Geothink&Learn video conference session on the topic of governing artificial intelligence. It highlighted Geothink’s unique interdisciplinary perspective and included a myriad of ideas from our faculty, students and partners. (Catch a recording of this event along with a transcript of the written chat that took place during the session below.)

The convener for the session was Geothink Head Renee Sieber, associate professor in McGill University’s Department of Geography and School of Environment. Speakers included Pamela Robinson, associate professor in Ryerson University’s School of Urban and Regional Planning and the associate dean for Graduate Studies and Strategic Services; Fenwick McKelvey, an assistant professor in Concorida University’s Department of Communication Studies; and, Elizabeth Judge, a professor of law at the Faculty of Law at the University of Ottawa.

A question and answer session followed after presentations concluded. Our four panelists briefly introduced their research and then reflected on the role of artificial intelligence in governance.


12:00:57	 From Pamela Robinson : Hello everyone! Thanks for joining us.
12:04:50	 From Geothink .CA  to  All Panelists : introduce Geohink
12:07:23	 From Geothink .CA : everyone loves nachos!
12:07:49	 From Patrick Connolly  to  All Panelists : it's true!
12:11:20	 From Geothink .CA  to  All Panelists : 2 min warning Renee
12:12:19	 From Geothink .CA  to  All Panelists : 1 min warning Renee
12:13:31	 From Geothink .CA  to  All Panelists : Time limit Renee
12:16:06	 From Patrick Connolly  to  All Panelists : that was great!
12:16:24	 From Renee Sieber : thanks!
12:16:46	 From Geothink .CA  to  All Panelists : Renee, are you eating nachos 🙂
12:17:06	 From Renee Sieber  to  All Panelists : scones—wish it was nachos
12:17:12	 From Pamela Robinson : If Zoom could also feed us it might be perfect
12:17:29	 From Geothink .CA  to  All Panelists : *Lunch&Learn
12:19:43	 From Patrick Connolly  to  All Panelists : as someone from toronto really interested in how google thinks and speaks in public about AI: if you happen to have a good reference or source (or keyword that I could use) for the comment on "google thinks ai will just be for small decisions" (paraphrased), then I'm totally interested 🙂
12:20:07	 From Patrick Connolly  to  All Panelists : for Renee ^^^ sorry
12:20:20	 From Pamela Robinson  to  Patrick Connolly, All Panelists : hi Pat - so nice you could join us 🙂
12:20:47	 From Pamela Robinson  to  Patrick Connolly, All Panelists : “Seeing” you twice in one week!
12:22:27	 From Geothink .CA  to  All Panelists : 2 mn warning elizabeth
12:23:01	 From Geothink .CA  to  All Panelists : 1 min warning elizabeth
12:23:11	 From Geothink .CA  to  Renee Sieber (Privately) : you may want to keep a tally of questions coming in through chat (so far 1 from Patrick Connley)
12:23:17	 From Renee Sieber : argument made by Google in a UK hearing on algorithmic transparency (http://data.parliament.uk/writtenevidence/committeeevidence.svc/evidencedocument/science-and-technology-committee/algorithms-in-decisionmaking/written/71681.html
12:23:39	 From Geothink .CA  to  Renee Sieber (Privately) : for the Q&A at end
12:24:00	 From Geothink .CA  to  All Panelists : Time limit warning Elizabeth
12:25:14	 From Fenwick McKelvey  to  All Panelists : Elections and AI is a very important subject area. Elizabeth Dubois and I have been working on bots as I have written about here: http://policyoptions.irpp.org/magazines/november-2017/toward-the-responsible-use-of-bots-in-politics/
12:26:50	 From Ricardo Hernandez : Several philosophers have tried to find or designed a logical structure to explain the moral and ethical issues and deducting general and acceptable outcomes, e.g. Kant`s moral imperatives, it is possible think in codifying this logical structures in algorithms in the AI case?
12:27:37	 From Geothink .CA  to  Renee Sieber (Privately) : ps - we have a few prospective partners on the call (Linda Low, van, city of winnipeg
12:27:53	 From Geothink .CA  to  Pamela Robinson (Privately) : ps - we have a few prospective partners on the call (Linda Low, van, city of winnipeg
12:28:47	 From Pamela Robinson  to  Geothink .CA (Privately) : ah cool - we’ll I am going to be super practical in mine so hopefully we’ll capture them between us all
12:29:06	 From Patrick Connolly : first pass at mapping quayside project, but some AI related branches in there that could perhaps be pulled out to make it more legible 🙂

12:30:06	 From Patrick Connolly  to  All Panelists : Pamela: *waves*
12:30:14	 From Patrick Connolly  to  All Panelists : Renee: thanks!
12:31:16	 From Geothink .CA : Great question, Ricardo - how to ‘codify’ ethics and the promises and threats embedded within
12:31:37	 From Rachel Bloom  to  All Panelists : How do I access the draft of that report by Treasury Board of Canada?
12:32:40	 From Geothink .CA  to  All Panelists : 2 min warning Fenwick
12:33:04	 From Pamela Robinson : Hi Rachel - good overview piece about it: https://medium.com/code-for-canada/responsible-ai-in-the-government-of-canada-a-sneak-peek-973727477bdf
12:33:45	 From Geothink .CA  to  All Panelists : 1 min warning Fenwick
12:35:06	 From Rachel Bloom  to  All Panelists : thank you!
12:35:11	 From Geothink .CA  to  All Panelists : Time limit reached Fenwick
12:37:28	 From Fenwick McKelvey  to  All Panelists : Michael Kaplan is taking the lead at the Treasury Board: https://medium.com/code-for-canada/responsible-ai-in-the-government-of-canada-a-sneak-peek-973727477bdf
12:38:47	 From Geothink .CA : Fenwick, McKelvey & Elizabeth, Dubois political bots report: http://comprop.oii.ox.ac.uk/wp-content/uploads/sites/89/2017/06/Comprop-Canada.pdf
12:39:05	 From Fenwick McKelvey  to  All Panelists : Here is the Treasury Board document: https://docs.google.com/document/d/1Sn-qBZUXEUG4dVk909eSg5qvfbpNlRhzIefWPtBwbxY/edit
12:39:20	 From Geothink .CA : Thanks!
12:41:10	 From Geothink .CA  to  Pamela Robinson (Privately) : start video
12:41:19	 From Geothink .CA  to  All Panelists : pamela, start your video again pls
12:42:16	 From Geothink .CA  to  All Panelists : 2 min warning Pamela
12:42:17	 From Geothink .CA  to  All Panelists : never mind Pamela, my machine is wigging out your video is on
12:43:14	 From Geothink .CA  to  All Panelists : is everyone ok with sharing their slides on our website?
12:43:24	 From Geothink .CA  to  All Panelists : 1 min warning pamela
12:43:50	 From Renee Sieber  to  Geothink .CA (Privately) : I have to put acknowledgements on my images
12:44:20	 From Geothink .CA  to  All Panelists : Time limit reached Pamela
12:44:33	 From Geothink .CA  to  All Panelists : that’s fine - I’ll just be reminding viewers at the end of the panel - doesn’t have to be right away
12:44:47	 From Bram Abramson : really interesting to think about atomized municipalities vs global platforms.  I wonder whether the Federation of Cdn Municipalities has done anything on this?  Assuming that their resources are limited — maybe they’d be an interesting partner for some of this owrk?
12:44:57	 From Fenwick McKelvey : Here is the Treasury Board document: https://docs.google.com/document/d/1Sn-qBZUXEUG4dVk909eSg5qvfbpNlRhzIefWPtBwbxY/edit
12:45:36	 From Pamela Robinson : A bunch of us from geothink.ca wrote for URISA journal about open data and infomediaries - you can find them here: 
http://www.urisa.org/clientuploads/directory/Documents/Journal/Vol28_final.pdf
12:46:14	 From Patrick Connolly : haha
12:46:18	 From Pamela Robinson  to  Bram Abramson, All Panelists : Bram - good question - yes FCM should be an important partner in this work.
12:47:09	 From Fenwick McKelvey : I’d like t share my college Joanna Redden’s great piece discussing risks of algorithmic governance: https://theconversation.com/six-ways-and-counting-that-big-data-systems-are-harming-society-88660
12:47:49	 From Geothink .CA  to  All Panelists : Pamela - question for you coming in
12:47:51	 From Geothink .CA  to  All Panelists : Pamela Robinson-- you talk about public consent in municipal AI projects.  At what point should that be taken into account?  Is it ethical for tech companies and/or governments to test technologies without this consent?  particularly if the information from this testing will be taken into account by any future algorithms..
12:47:58	 From Renee Sieber  to  Geothink .CA (Privately) : We have a question in the Q&A section: you talk about public consent in municipal AI projects.  At what point should that be taken into account?  Is it ethical for tech companies and/or governments to test technologies without this consent?  particularly if the information from this testing will be taken into account by any future algorithms.
12:48:10	 From Geothink .CA  to  All Panelists : ill read it out to Pamela
12:48:25	 From Renee Sieber  to  Geothink .CA (Privately) : Hello, this is a question for Fenwick McKelvey. Can you expand on net neutrality and AI? Thanks. Maroussia
12:48:57	 From Geothink .CA  to  All Panelists : question for Fenwick
12:48:58	 From Geothink .CA  to  All Panelists : Hello, this is a question for Fenwick McKelvey. Can you expand on net neutrality and AI? Thanks. Maroussia
12:49:08	 From Geothink .CA  to  All Panelists : I’ll read it out
12:49:30	 From Tenille Brown  to  All Panelists : I am sorry I have to get going. Thank you for your presentations. I really enjoyed the information.
12:49:57	 From Renee Sieber  to  Tenille Brown, All Panelists : Thanks, Tenille!
12:50:04	 From Fenwick McKelvey : Have a listen to: http://www.wnyc.org/story/32-ok-cupid/
12:50:10	 From Patrick Connolly : if we're thinking of AI as an "abstraction" to make the complex simpler, of which we have many in government and society, is there a precedent (for or against) making abstractions legible or interrogable when they're related to collective decision-making?
12:50:37	 From Tenille Brown  to  All Panelists : all of these URLs are now open in my browser for me to look at later. Thanks
12:51:06	 From Geothink .CA : Viewers - feel free to ask written questions on the Q&A portal top left of your screens 🙂
12:51:10	 From Renee Sieber  to  Geothink .CA (Privately) : Possibly for Elizabeth: Several philosophers have tried to find or designed a logical structure to explain the moral and ethical issues and deducting general and acceptable outcomes, e.g. Kant`s moral imperatives, it is possible think in codifying this logical structures in algorithms in the AI case?
12:51:26	 From Geothink .CA  to  Renee Sieber (Privately) : got it
12:51:35	 From Patrick Connolly : ah thanks! wasn't obvious on mobile 🙂
12:51:41	 From Geothink .CA  to  Elizabeth Judge (Privately) : possible Q for you Elizabeth
12:51:43	 From Geothink .CA  to  Elizabeth Judge (Privately) : Possibly for Elizabeth: Several philosophers have tried to find or designed a logical structure to explain the moral and ethical issues and deducting general and acceptable outcomes, e.g. Kant`s moral imperatives, it is possible think in codifying this logical structures in algorithms in the AI case?
12:52:45	 From Elizabeth Judge  to  All Panelists : ok
12:53:16	 From Pamela Robinson  to  Geothink .CA (Privately) : Why is my name up there?
12:53:26	 From Pamela Robinson  to  Geothink .CA (Privately) : can you not see me?
12:54:02	 From Geothink .CA  to  Pamela Robinson (Privately) : no idea - are you logged in elsewhere? does Lisa Ward have your login, I swear I saw her name earlier…
12:54:04	 From Geothink .CA  to  Pamela Robinson (Privately) : I can see you
12:54:13	 From Geothink .CA  to  Pamela Robinson (Privately) : but there is a second box with only your name
12:54:16	 From Pamela Robinson  to  Geothink .CA (Privately) : She might
12:54:18	 From Pamela Robinson  to  Geothink .CA (Privately) : that’s ok
12:54:27	 From Geothink .CA  to  Pamela Robinson (Privately) : yeah - i blocked the video
12:55:49	 From Geothink .CA : Viewers - feel free to press ‘Raise Your Hand’ to record a live video question for our panelists
12:57:42	 From Renee Sieber  to  Geothink .CA (Privately) : From Patrick: If we're thinking of AI as an "abstraction" to make the complex more simple (of which we have many in government and society), is there a precedent, either for or against, to asking for these abstractions to be made legible or interrogable when they're related to collective decision-making?
12:57:53	 From Patrick Connolly : FYI Fenwick, ppl in polis and g0v slacks were really excited about this tweet you were featured in 😉

12:58:08	 From Geothink .CA  to  Renee Sieber (Privately) : *not yet 😉
12:59:05	 From Geothink .CA  to  Renee Sieber (Privately) : got it renee
12:59:28	 From Geothink .CA  to  Renee Sieber (Privately) : was saying not yet on ‘nothing in Canada on AI”
12:59:28	 From Patrick Connolly : and I quote:

> @patcon for all we try to move you to nyc, canada gov is far ahead of us. exciting news stirring
13:00:07	 From Rob Davidson : Thanks Fenwick and Pamela for bringing existing intervention and advocacy models to note in the AI context.
13:00:48	 From Fenwick McKelvey : Thanks Patrick!
13:01:19	 From Pamela Robinson : Rob and everyone else, we all need to put our best heads on answering the question: what models have worked and have can we adapt them to this context?
13:02:47	 From Geothink .CA : Thanks, Rob
13:03:11	 From Geothink .CA  to  Renee Sieber (Privately) : sorry - thought Elizabeth would have feedback
13:03:42	 From Renee Sieber  to  Geothink .CA (Privately) : you can ask Elizabeth directly to respond
13:03:45	 From Renee Sieber  to  Geothink .CA (Privately) : Also Several philosophers have tried to find or designed a logical structure to explain the moral and ethical issues and deducting general and acceptable outcomes, e.g. Kant`s moral imperatives, it is possible think in codifying this logical structures in algorithms in the AI case?
13:04:29	 From Geothink .CA : Great point, Fenwick - for e.g. why should we move away from using ‘Lena’ standard in training datasets for AI
13:04:34	 From Renee Sieber  to  Geothink .CA (Privately) : what time do we end?
13:04:35	 From Fenwick McKelvey : Good example of bias: https://www.wired.com/story/machines-taught-by-photos-learn-a-sexist-view-of-women/
13:04:57	 From Geothink .CA  to  All Panelists : reminder - we end at 1:15 today
13:05:05	 From Patrick Connolly : re:that tweet. aaaand I feel the panelists here would make the folks in g0v/polis chats even more excited. this has been so rad!
13:05:26	 From Geothink .CA : Agreed, Patrick!
13:05:45	 From Geothink .CA : Great example, Fenwick - unfortunately, too many…
13:06:25	 From Geothink .CA  to  Renee Sieber (Privately) : standards! 🙂
13:10:10	 From Patrick Connolly : "coherent blended volition" is one theoretical way I've heard to blend human and machine intuitions
13:10:14	 From Geothink .CA  to  All Panelists : *I’ll be telling viewers power points will be made available on our website - objections?
13:10:14	 From Fenwick McKelvey : Here is a good web game where you can pretend to train a self-driving car: http://moralmachine.mit.edu/
13:10:26	 From Diogo Pereira Henriques  to  All Panelists : Apologies, I only arrived in the discussion. I guess the need of legislation is an important issue of all ML and AI utilised in public services/infrastructures...
13:10:37	 From Patrick Connolly : https://wiki.lesswrong.com/wiki/Coherent_Blended_Volition
13:10:57	 From Diogo Pereira Henriques  to  All Panelists : Thank you for organizing this discussion !
All the best,
13:11:14	 From Patrick Connolly : architects of ai-assisted vtaiwan process were thinking about it
13:11:17	 From Patrick Connolly : https://blog.pol.is/uber-responds-to-vtaiwans-coherent-blended-volition-3e9b75102b9b
13:11:27	 From Renee Sieber  to  Diogo Pereira Henriques, All Panelists : We will put the webinar online by the end of next week
13:11:37	 From Ricardo Hernandez : thank you prof. Judge for your answer
13:12:16	 From Geothink .CA : Fenwick, does the game also include compliance, insurance, safety, etc 🙂
13:13:14	 From Geothink .CA : Folks, we’ll be finishing in a few minutes - feel free to post final remarks or questions 🙂
13:14:54	 From Geothink .CA : Thanks to all attendees for such an engaged convo
13:16:13	 From Patrick Connolly : with chat log pls!
13:16:23	 From Patrick Connolly : thanks everyone!
13:16:49	 From Diogo Pereira Henriques  to  All Panelists : Thank you
13:16:49	 From Fenwick McKelvey : Thanks so much. Its been a real pleasure

Download a PDF of this event to share.

When:
Thursday, December 7, 2017 at 12:00 PM [NOW CONCLUDED]

Where:
https://zoom.us/webinar/register/WN_36jBhV9ZTea4n19hiCPW1w

After registering, you will receive a confirmation email containing information about joining the webinar.

Convener:
Renee Sieber

Moderator:
Sonja Solomun

Panelists:

Renee Sieber: How do we hold AI accountable, especially when we don’t know how AI works?

 

 

 

 

Pamela Robinson: So now, as a planner, I have to invite computers to my public meetings?!? Early thinking on the implications of emerging AI on the planning and design of cities 

 

 

 

Fenwick McKelvey: What should broadcasting teach us about AI policy?

 

 

 

 

Elizabeth JudgeElizabeth Judge: Smart Cities, Smart Driving: Artificial Intelligence, Automated Decisionmaking, and the Ethics of Transportation.