Scenario Building Email Excerpts - mitmedialab/2018-MIT-IAP-ComputationalLaw GitHub Wiki

The following email excerpts were taken from correspondence in preparation for building scenarios to be used as discussion anchors at the 2017 MIT Legal Forum.

Watson-Related AI Scenario Building

... rough out a short anchor scenario identifying a fundamental area of law and describing a relevant use technical case and/or business case and/or legal fact pattern for combination of those that you can use to anchor discussion in a way everybody can share a common reference point back to in your session as well as across all breakout group sessions.

So that I’m not just saying “do something” without context of what to do or how to do it or examples of successful states of “done” here are a couple of ideas:

One great scenario would be in the area of regulatory compliance and or litigation avoidance, or perhaps roll all that into a higher level scenario of risk management and avoidance. That would also be a great canvas for introducing you to my contacts in the GE legal team who are quite focused on this and ways to use computation to rise above their current ceilings and levels of performance. Watson is computational and I can think of some ways to tie in various Watson capabilities to extremely cool data they have and processes they are working on.

Also, I recall you saying something about public law and regulation as a key use case in the past and there is a scenario document in our shared Google Drive under the scenarios folder at the root level named public sector scenario for AI and blockchain. I would hate to see the public sector scenario go on explored for lack of anybody picking it up and I fear I may not have time to do it by myself, however if you wanted to do your lightning talk on that scenario I could definitely help a lot to fill it out and include appropriate references and thought-provoking questions and so forth.

I encourage you to look at the files in the scenarios folder to get a sense of how people are approaching this and to take great comfort in the astonishing variety of approach, level of detail and style. The only criteria I feel is vital for a scenario to serve as a usable discussion anchor for purposes of what we are trying to do in this legal for him event is that you clearly identify the key people including the name or names of their most relevant role or roles and just is clearly identify the key actions or interactions or transactions they are doing. The ideal level of detail would focus on two or three actors and one or two key actions that spotlight the most important prevailing context for the scenario and do the best job of surfacing the dynamics and the points you want to bring out for your topic. For example:

Actors: ACME Corp INC (role: Product manufacturer and UCC 2 merchant) and Alice Smith (B: Customer; L: Consumer; T: User)
Actions: 1) Alice purchases product from ACME. 2. ALICE and ACME agree to product diagnostic data collection. 3. ACME does network upgrade to product to fix dangerous problem.
Fact Pattern: Consumer(s) and ACME Corp Inc agree to anonymous product diagnostic reporting from ACME Product purchased by consumer. Aggregate and anonymized data when analyzed with analysis and analytics (ie: slot in the magic tech here) recognizes a pattern of anomalous behavior and product failure that triggers trouble shooting and reveals a previously unknown product failure condition that could be dangerous.
Legal Result: ACME safeguards customers, preserves brand integrity and avoids probable class action lawsuit by fixing the product failure via firmware upgrades and notices to potentially effected users with info on how to avoid harm by addressing the possible issue if it happened or happens to them.
Breakout Group Suggested Discussion Points: What legal and marketing frameworks would be needed to enable this use of beneficial AI and to ensure trustworthy consumer sources mass-scale data collection and periodic product diagnostic “phone home” reporting?
note to facilitator: mention the emerging best practice standards of GDPR and the “New Deal on Data” approach whereby data reporting method uphold data protection and respect individual identity to enable legitimate AND very timely/valuable/actionable results at key junctures in the lives of individuals and also at population-scale.

AI Sandbox Related Scenario Building

I've thought more and I like the idea of structuring a breakout around AI Sandbox. As a way to focus on a common, understandable discussion anchor, the format of the breakouts have been designed to anchor discussion on a legal scenario rather than on a computational tool or other technology implementation.

I think I'd need to see a demo to understand better and perhaps make more astute suggestions, but off the top of my head I think the best course would be to select the most likely (strongest candidate) near-term scenario whereby tools such as those of AI Sandbox materially transform a field of law, area of law practice or type of legal process. Taking AI Sandbox as demonstrating a wide range of AI technology, tools and techniques in the legal field we can postulate the computational law offices of a near-term future with integrated, automated information flows among all relevant parties enabling AI-driven legal offices. Fleshing out a couple paragraphs the evocatively describe the "common" near-term future practice of computational law in some context. By sketching out an understood existing area of law practice there is more room for people of all experience and expertise levels who are familiar with law OR familiar with AI to jump in the dialog and fruitfully contribute.

If there is an area of law practice you already know to be evolving and you can guess with some confidence the shape of a future stable state, then that would be the best context to extrapolate. If you are not especially confident then I'd suggest considering an immigration practice (thanks to some ringer small office entrepreneurs who will be present at the conference) or a bankruptcy practice or a mass torts/class actions practice.

The final thing would be to write up some questions/issues and some options/opportunities to help catalyze idea flow and good group discussion. The sweet spot, in my opinion, for catalyzing questions or other interventions is to spark discussion that sharpens the mind on and illuminates the path toward refactoring and transforming law and legal processes to digital, networked and computational rules, processes and systems.

... Also...I'd add that in the context of a fully automated law office described in my immediately prior email, imagine all that systemic processing and data integration enabled legal advice that was so much more accurate, timely, efficient and effective that it was impossible to compete against by practitioners who lacked computational law practice systems of comparable sophistication and performance. In this way one can imagine a transformation of that given area of legal practice.

Now imagine that the AI super-charging of client advice was itself derived from data-driven situational assessments, objective strategic benchmarks and continuous tactical predictions resulting in multivariate priority recommendations fed into algorithmic client-interest-optimization models. Open questions about impacts of AI-infused law and legal processes include...

How do the needs, capabilities and roles of clients and attorneys evolve and how does that evolution transform the nature of the lawyer-client relationship?
What are the current best practices and most promising proposed future safeguards for a fiduciary to identify, measure and adjust for the various biases, assumptions, priorities and goals infused invisibly in the technology?
When, where and in what form are legal services accessed once law and legal processes are refactored into digital information flowing over connected networks?
Do legal processes such as signatures and notarization, contracts and licenses, or authorization and content management become another form of JSON object that can express itself as a service? processed through highly dimensional and adaptive rules based system?
This scenario should also be informed by the current best practices and proposed future safeguards for identifying, measuring and adjusting for the various ways biases, assumptions, priorities and goals become invisibly infused in the technology.