Below is an overview of the current discussion topics within the Trusted AI Committee. Further updates will follow as the committee work develops.
Focus of the committee is on policies, guidelines, tooling and use cases by industry
Survey and contact current open source Trusted AI related projects to join LF AI & Data efforts
Create a badging or certification process for open source projects that meet the Trusted AI policies/guidelines defined by LF AI & Data
Create a document that describes the basic concepts and definitions in relation to Trusted AI and also aims to standardize the vocabulary/terminology
Please self subscribe to the mail list here at https://lists.lfaidata.foundation/g/trustedai-committee.
Or email email@example.com for more information.
Trusted AI Committee North America Monthly Meeting - 4th Thursday of the month, 10 PM Shenzen China, 4 PM Paris, 10 AM ET, 7 AM PT USA (updated for daylight savings time as needed)
Zoom info : https://zoom.us/j/7659717866
We will starting ONE Monthly Call for Europe/Asia Time Zone, every 2nd Thursday of the month (Starting in November - time to be determined)
Initial Organizations Participating: IBM, Orange, AT&T, Amdocs, Ericsson, TechM, Tencent
Responsible Innovation Project
The Institute for Ethical AI and Machine Learning
sumalaika (but different email address)
- All the assets being
- Fairness: Methods to detect and mitigate bias in datasets and models, including bias against known protected populations
- Robustness: Methods to detect alterations/tampering with datasets and models, including alterations from known adversarial attacks
- Explainability: Methods to enhance understandability/interpretability by persona/roles in process of AI model outcomes/decision recommendations, including ranking and debating results/decision options
- Lineage: Methods to ensure provenance of datasets and AI models, including reproducibility of generated datasets and AI models
AI Fairness 360
Adversarial Robustness 360
AI Explainability 360
Trusted AI Principles Working Group
Trusted AI Technical Working Group
Zoom info : Trusted AI Committee meeting - NA/Europe Meeting 4th Thursday of the month, 10 PM Shenzen China, 4 PM Paris, 10 AM ET, 7 AM PT USA (updated for daylight savings time as needed)
How to Join: Visit the Trusted AI Committee Group Calendar to self subscribe to meetings.
Or email firstname.lastname@example.org for more information.
Meeting Content (minutes / recording / slides / other):
|November 19, 2020|
|October 1, 2020|
Notes from the call:
Notes from the call:
Notes from call:
Ofer, Alka, Francois, Nat, Han, Animesh, Jim, Maureen, Susan, Alejandro
Notes from the call:
Animesh Singh (IBM), Maureen McElaney (IBM), Han Xiao (Tencent), Alejandro Saucedo, Mikael Anneroth (Eriksson), Ofer Hermoni (Amdocs)
Animesh will check with Souad Ouali to ensure Orange wants to lead the Principles working group and host regular meetings. Committee members on the call were not included in the email chains that occurred so we need to confirm who is in charge and how communication will occur.
The Technical working group has made progress but nothing concrete to report.
A possible third working group could form around AI Standards.
Notes from the call:
Notes from call: https://github.com/lfai/trusted-ai/blob/master/committee-meeting-notes/notes-20190919.md
Here are some additional notes taken by Jim Spohrer (IBM) <email@example.com> - https://docs.google.com/document/d/1QJ3t0YD3mzOa-gajjbHtgjXEO7_LPY1gmRW1SHcoNo4/edit
This document: https://docs.google.com/document/d/1QJ3t0YD3mzOa-gajjbHtgjXEO7_LPY1gmRW1SHcoNo4/edit?usp=sharing
LF AI - Trusted AI Committee Call - Every other week
Ibrahim Haddad (LF): I will set up "Trusted AI" Zoom to record future calls
Ibrahim: Attendees should go to Wiki for agenda: https://wiki.lfai.foundation/display/DL/LF+AI+Trusted+AI+Committee
Jim Spohrer (IBM): We should rename ourselves when we join zoom - so people can easily see your company - I just did that (a good point to add to meeting start when we review the agenda)
Nat Subramanian (Tech Mahindra): Wiki todo: Need a section of wiki to highlight guest speakers
Ibrahim: Zoom todo: will set up a zoom for Trusted AI - and in future co-chairs will be able to record
Animesh Singh (IBM): Showed a Google Doc with agenda - need to get link
Animesh: Two working groups (WG) will have lead, core members, optional members and meet separately
Animesh: WG Principles and WG Technical
Animesh: WG Techincal focus - landscape projects in Trusted AI area, and specifically those that are coming into LF AI to support Trusted AI work
Animesh: Which OS projects will come into LF AI and which are being integrated into LF AI projects
Animesh: Face-to-face meetings for WG will need to be planned by WG leads
Animesh: WG should have outcome and timetable
Animesh: sharing a Google Doc with the above information
Maureen McElaney (IBM): I’ve added a loose agenda for today’s call to the wiki. Since I’m on mobile can someone take on the responsibility of adding any further notes there?
Jim Spohrer (IBM): I am taking notes and can clean them up and upload them, link from Wiki to notes
Animesh: Introducing Romeo Kienzler (IBM Switzerland) who added LF AI meeting in Paris hosted by Orange
Animesh: Romeo has created top data sciences and AI Coursera courses
Alka Roy (AT&T Innovation Center) - 7:15 AM PT - Had trouble with new link - just joining.
Romeo Kienzler (IBM Switzerland): Working on pipelines for technical working group
Animesh: Can we use Slack to keep an ongoing work cadence ?
Nat: Not slack because of licensing complications. ICRC is being used instead. Communication channel gap.
Nat: Acumos has SCRUM meetings almost everyday
Nat: Acumos meetings - we can re-use time slots and leverage for Trusted AI WG.
Nat: Tencent Angel team interest in on boarding a model
Nat: two working groups - Principles and Technical
Nat: Synergies between Acumos + Angel + AIF360
Nat: Principles works on policies and strategies
Alka Roy (AT&T Innovation Center) - 7:30 AM PT: Good morning!
Alka: How to get to a set of criteria and guidelines?
Animesh: job of WG Principles
Alka: Principles team creates, and technical team provides feedback/adopts
Animesh: yes, criteria evolves with feedback from terchnical
Nat Subramanian (Tech Mahindra):: Romeo is IBM in Switzerland?
Nat: China has most difficulty joining with times we have proposed so far (early California)
Animesh: Romeo’s time might be better for China and India meetings
Animesh: Let’s identify right subset of people, and let’ them find a good weekly time for them; Since Nat knows the folks - he can best find a way to loop in Romeo. I am OK with calls after 10pm PT
Nat: I will get a time for Technical WG
Animesh: I will add Romeo email to WIki - his email is firstname.lastname@example.org
Francois Jezequel (Orange) - 7:32 AM PT: Suoad is willing to continue, just not available today
Francois: “Sue-wad” rhymes with “Squad”
Alka: IEEE has good work we can borrow from
Alka: how to apply principles, practical technical and principle WG interactions
Alka: going beyond theory into practice will be helpful
Animesh: Required people and optional people
Animesh: Lineage -> Certification -> Badging
Animesh: Can help connect with IBM Research and ATT and rest of Principles committee
Animesh: Factsheets project is relevant - Factsheets: Increasing Trust in AI Services Through Supplier Documents of Conformity - Certifications - https://arxiv.org/pdf/1808.07261.pdf https://www.ibm.com/blogs/research/2018/08/factsheets-ai/
Alka: Please add me to the Principles (optional), and I will find the right ATT required
Animesh: Alka is Bay Area for a visit sometime to discuss
Alka: Rueben (AT&T)
Alka: I will make sure AT&T is engaged
Nat: I have met with Rueben in NJ
Jeff Cao (Tencent) - 7:44 AM PT: I have ideas for person on technical person - Fitz Wang (Tencent)
Nat: I will reach out to Fitz, and please share name of others
Mikael Anneroth (Ericsson) - 7:45 AM PT: please add Ericsson to Principles work group - would like to share and discuss - I am in Stockholm Sweden
Jeff Cao (Tencent): Han Xiao is on the list
Animesh: Susan Malaika (IBM) is also interested)
Animesh: If Orange will drive Principles
Jeff Cao (Tencent): Tencent can join the principles group
Francois Jezequel (Orange): Yes, Souad is interested in leading Principles WG
Alka: identify gaps since a lot of work has already bee done
Francois: Minimum intersection that can be operationalized (how to make it practical in operation)
Alka: Please let’s make the goal in writing, and when to have meeting to share each perspective, consolidate, and agree
Animesh: Francois or Souad can send a kick-off email, and we can get each group add their members (like IBM include Susan, Maureen, Jim, etc.
Animesh: Shared several Google doc links and will upload them to Wiki
Alka: Strategy meeting - f2f being planned - in October - by that time converged on principles - progress to report
Ofer: End of October, OSS Oct 31, full day governing board meeting
Nat: details readout in Oct 31
Alka: good deadline
Alka: Trusted and Responsible AI added to mission for LF AI - any objections
All: No objections noted
Working Group Names and Leads:
Principles, lead: Souad Ouali (Orange France) with members from Orange, AT&T, Tech Mahindra, Tencent, IBM, Ericsson, Amdocs.
Technical, lead: Romeo Kienzler (IBM Switzerland) with members from IBM, AT&T, Tech Mahindra, Tencent, Ericsson, Amdocs, Orange.
Working groups will have a weekly meeting to make progress. First read out to LF AI governing board will be Oct 31 in Lyon France.
The Principles team will study the existing material from companies, governments, and professional associations (IEEE), and some up with set that can be shared with the technical team for feedback as a first step. We need to identify and compile the existing materials.
The Technical team is working on Acuomos+Angel+AIF360 integration demonstration.
Thanks Souad for your note below to committee members - just re-reading it...
(1) Great to keep a global balance on the committee
(2) Look forward to definitions you are drafting
(3) Collecting material - see some links below.
(4) Cross analysis - Harvard study good starting point.
(5) Pragmatic approach - what real-world enterprise use cases?
(6) Open source tools - how tools and Trusted AI Workflows relate to above use cases?
(7) Watch on LFAI landscape projects - which projects have been checked against principles?
(8) How to ensure trust in open source package - how to organize audit to ensure files not corrupted? See Factsheets link below for a start.
Some useful link:
Globally Diverse - LFAI Trusted AI Committee: https://wiki.lfai.foundation/display/DL/Trusted+AI+Committee
University: Harvard AI Principles meta-analysis: 32 sets of principles side by side, with 8 themes: https://ai-hr.cyber.harvard.edu/primp-viz.htm
Company: AT&T (Tom Moore, Chief Privacy Officer): https://about.att.com/innovationblog/2019/05/our_guiding_principles.html
Company: IBM Everyday Ethics for AI: https://www.ibm.com/watson/assets/duo/pdf/everydayethics.pdf
Company: IBM NIST AI Standards response: https://www.nist.gov/sites/default/files/documents/2019/06/06/nist-ai-rfi-ibm-001.pdf
Company: IBM Factsheets: Increasing Trust in AI Services Through Supplier Documents of Conformity - Certifications - https://arxiv.org/pdf/1808.07261.pdf
Company: Tencent sent some files - but I do not have a URL to the documents.
Trusted AI group calendar: https://wiki.lfai.foundation/pages/viewpage.action?pageId=12091895 (Please subscribe)
Mailing list: https://lists.lfai.foundation/g/trustedai-committee (Please subscribe)
Q: Where is the best place for us to upload documents to share with the Trusted AI Committee/Principles WG?
Some notes from Oct 17, 2019 committee call:
- Animesh walked through the draft slides
(to be presented in Lyon to LFAI governing board about TAIC)
- Discussion of changes to make
- Discussion of members, processes, and schedules
- Jim will put slides in Google Doc and share with all participants
- Susan is exploring a slack channel for communications
- Trust and Responsibility, Color, Icons to add Amdocs, Alejandro's Institute
+ @Ofer to PWG
Alka: helped draft these - and acknowledges need to get everyone’s perspective - asked about feedback loop with Usecases Group
Francois: These slides will be a good resource - thank-you; now we have to organize the way we contribute to each part of the document - Suoad will lead getting consensus and finalizing this draft
Francois: Still need to define connection between principles and tooling - and make it concrete Animesh: Members of UWG? Animesh: Need members who will create use cases - software and actual industry use cases Animesh: What are the telco use cases?
Alka: personal focus on PWG definition. Will work to gather AT&T use cases and try to find the right active person
Nat: interest yes. However, current focus on the release - in 3-4 weeks from now release will be complete for Acumos. Technical people tied up on release. Nat: Reuben can give guidance, and Nat working on next level of technical experts
Han: Jeff and I working to get more awareness and attract others in Tencent; open source office and develops team to get involved. Find real world use cases
Francois: Discussion of when PWG will meet - off-line and maybe two meetings. (Alka, Francois, Jeff, Susan)
Alka: I have material to share with everyone when we can schedule that
Animesh: Alejandro has great material on GitHub.
January 23 meeting notes:
Attendees: Animesh Singh (IBM), Samuel Hoffman (IBM), Nat (TechM), Susan Malaika (IBM), Alejandro Saucedo (Ethical AI Institute and Seldon), Maureen McElaney (IBM, Ofer Hermoni, Parag Ved, Yassi Moghaddam (ISSIP), Jim Spohrer (IBM)
Animesh: Asked Nat for Acumos update
Nat: Acumos update
Animesh: Introduced Samuel Hoffman (IBM Research) to give Trusted AI and Scikitlearn demo
Nat: Will the IBM materials be donated to LF AI? Aligned with IBM Product OpenScale?
Animesh: Contributed if LF AI pull and community adoption
Jim: Correct, product team at IBM wants to see community pull for it to be contributed (actual use and contributions from community)
Nat: Mentioned Acumos Gaya - and will send links
Ofer: Introduce Yassi yourself
Yassi: introduced herself as exec director of professional association (Jim on Board, Susan lead on AI speaker series)
Nat: gave history of the LF AI Trusted AI Committeee - companies from Asia, USA, Europe involved - two working groups, use cases and principles
Nat: constantly evolving area
Ofer: discuss it