Week 7–What are the Regulatory AI Projects?
After weeks of promises, I finally explain in detail the two AI projects I am working on!
Before I explain what the two Regulatory AI projects are, I wanted to make an appeal to those in Ottawa who have the time to volunteer and help fill sandbags. Many of our fellow residents are bracing for record level floods (thought to exceed the levels seen during the 2017 flooding). If you want more details on how to help you can visit: https://ottawa.ca/en/news/what-expect-if-youre-volunteering-flood-preparations
What is a community if we don’t help each other out during a time of need? We might be living in the darkest timeline but the way we make a difference is to help our neighbours and show empathy. If you can’t volunteer, there are other ways to contribute such as donating to disaster relief organizations.
We should measure our success in life based on how we make the world a better place; don’t stress so much about how many problems the world has but rather what small actions you can take today to make someone’s day better.
Regulatory Evaluation Platform Artificial Intelligence Project
The majority of this week was spent with after-work due to the bidder engagement session held for the Regulatory Evaluation Platform project. Receiving over 40 questions leads to a large amount of work to provide good answers.
So what is the Regulatory Evaluation Platform Artificial Intelligence Project (or REP for short)? REP is one of two projects which are part of the AI for Regulatory Insights demonstrator projects I am leading.
Regulators are required under the Cabinet Directive on Regulation to undertake an exercise known as “regulatory stock taking” a lifecycle approach to the analysis of regulations to ensure regulations are encouraging innovation and growth while protecting health, safety and environment, looking for outdated language, redundant regulations, potential linkages with regulations in other jurisdictions (provincial or foreign). It goes without saying that there is a level of complexity when are you dealing with thousands of regulations (2600+) at the Canadian federal level alone plus thousands at the provincial and international level.
Currently, analysts and drafters review each regulation for linkages and overlap manually searching related regulatory documents using internet/spreadsheets. A manual and labour intensive process that is prone to human error and would require a significant number of hours of labour to complete. There is no systematic tool for compiling info on multiple jurisdictions, by sector/industry, to evaluate complexity, flexibility, conflict, and terminology
The REP project will accelerate more complex analysis and new insights into regulations with a material impact on stakeholders. The use of natural language processing, semantic analysis and artificial intelligence will lead to timely and less labour intensive insights and analysis to inform advice or decisions. It will allow for:
- Cluster/network analytics for interjurisdictional burden of overlapping, obsolete or outdated regulations
- Linking acts/regulations to NAICs codes to enable sector studies combining multiple data/info sources
- Sentiment analysis and potential plugin to consultation data
The RFP is launched and live(with the support of Transport Canada, Public Services and Procurement Canada, Community of Federal Regulators, Canada School of Public Service, Environment and Climate Change Canada, Justice Canada, Agriculture Canada, Innovation Science and Economic Development Canada and Treasury Board Secretariat ).
Bids will be received by June, three firms will be selected to develop prototypes by October and a final product will be ready by March 2020.
In future weeknotes, I will dive deeper into why we are doing this project, how this relates to CSPS’ mandate, what we hope to learn and why our process is just as important as the output. However, these are topics that deserve their own space and today I wanted to explain in greater detail what these projects are. If you’re at the edge of your seat and can’t wait then hit me up at scott.mcnaughton@canada.ca
Incorporation by Reference AI Project
So what is the Incorporation by Reference AI Project (or IBR for short)? IBR is one of two projects which are part of the AI for Regulatory Insights demonstrator projects I am leading.
Incorporation by Reference is a common practice in the regulatory field where a regulation makes reference to another regulation, standard or other document so as to avoid having to “re-create” that document in the regulation itself. This practice offers many benefits to regulators. Primarily, by making a reference to another document in their regulations, regulators can avoid having to do regulatory amendments every time a standard is updated. This practice ensures regulators are always using the latest and greatest standards in their regulations.
However, there hasn’t been a proper “count” of the use of incorporation by reference in federal level regulations. The incorporation by reference identification and review process is currently time consuming and manual (estimated at over 4,000 references in regulations). As a result, significant staff time is spent trying to find and record attributes of documents incorporated by reference (e.g. location, cost, language). Finally, a manual process is at risk of not being current and so continuous monitoring becomes an ongoing and major challenge.
The IBR project aims to automate the process to improve the speed and accuracy of monitoring. This would allow for timely and accurate counts and provide insights into the location, cost and language access of documents which are incorporated by reference. Drafters and regulators may use that information when updating or revising their regulations.
It is estimated that the current manual process would take over 1,300 hours to complete and would need to be repeated on an annual basis. Automating this process would mean the same taking a few hours and could be completed on demand whether at an individual regulation level or across the entire regulatory stock.
We’ve engaged a contractor who is building the scraping tools, data visualization UI and algorithm which will support the automation of IBR. A prototype is expected by May 2019 and a production ready solution by August 2019.
As I said in the last section, I will dive deeper into why we are doing this and why it matters etc. in future weeknotes. If you’re at the edge of your seat and can’t wait then hit me up at scott.mcnaughton@canada.ca
Other Work — Rules as Code
I’ve talked about rules as code in past weeknotes. I am starting to branch out and talk to people about real use cases. There seems to be a lot of support and interest. From my angle, there is a lot to learn in this space for rule makers and given some of the new models the Canada School of Public Service is testing around procurement, interdepartmental collaboration, MOUs, risk taking etc. a solid use case is all that is needed to get work started. So if you are working on “Rules as Code” please reach out at scott.mcnaughton@canada.ca
What’s on my mind?
What’s on my mind? Get in touch if you want to chat or have an insight to share.
- The tweet I linked above where Hertz sued Accenture for failing to deliver on a $32 million “digital transformation” project. Anyone surprised? Has an outsourced digital transformation ever worked? Why do we fall into these same traps over and over?
- If the Earth is passing the point of no return for climate change in 12 years, why are humans not mobilizing en masse for a world extinction level event? Why is there no sense of urgency around anything anymore?
- Should people leading change spend time trying to convert cynics? Are cynics just “realistic” optimists? Are they cynical because they have been hearing the same “ra-ra” cheerleading change message for years over and over with no real observable change?
Week 7 of my weeknotes comes to an end. I hope my detailed notes on two of the projects I am working on are interesting and noteworthy. I’m always open for a coffee so hit me up!
