Slides for lecture 1 of the course Introduction to Legal Technology at the University of Turku Law School, presented Jan 13 2015.
This lecture contains an overall introduction to the course and presents some general legal tech/compsci/AI concepts.
Analysis on Law of Domicile under Private International laws.
Introduction to Legal Technology, lecture 1 (2015)
1. TLS0070 Introduction to
Legal Technology
Lecture 1
Introduction
University of Turku Law School 2015-01-13
Anna Ronkainen @ronkaine
anna.ronkainen@onomatics.com
2. Welcome!
...on a journey to the unknown.
(Really, I have no idea what will actually come out of this. Caveat auditor.)
3. So, what is this?
- not a course about the legal regulation of
technology (save for lecture 8)
- not your average legal theory course (even if the
teacher is a legal theorist...)
- it’s about technology (ICT) as it relates to
the practice of law
- existing technology but even more so future
technology, and the changes it will bring to
the legal profession – and your career
- 1st course of its kind in Nordics/Baltics (?)
5. Look who’s talking
Anna Ronkainen
- a lawyer at least on paper (LL.M., U of Cph);
also studied EE/CS, linguistics; researcher in
computational legal theory (U of Hki)
- Chief Scientist and co-founder, Onomatics Inc.
- worked in the software industry since the early
1990s
- my knowledge of practical lawyering comes from just very little in-
house work (and a lot of films and TV), so please correct me
whenever I’m wrong...
7. So, why this now?
- in 1995, lawyers didn’t use e-mail – and
were vehemently against the whole idea
- ... and now ...
- in 2035, who knows? (and if you’re in law school now, you
probably expect to practice well into the 2050s)
- what we do know is that a change is
underway – and where it’s coming from –
and that’s what this course is all about
- for example: legal startups have gone from
about 20 to about 500 in the past 5 years
9. Course format
- 10 lectures (2 h each) on Tuesdays between
14 and 18 (check the schedule!)
- attendance not mandatory but strongly
recommended (20% of grade)
- final paper: 2500–4000 words (80% of
grade)
- required reading: Richard Susskind:
Tomorrow’s Lawyers (OUP 2013) and some
articles (see Moodle)
10. Lesson plan (1/2)
1. Introduction. On law and technology. What is
legal technology? (Jan 13, 16–18)
2. Artificial intelligence and law: the 20th century
(Jan 27, 14–16)
3. Artificial intelligence and law: the 21st century
(Jan 27, 16–18)
4. Human factors: What does AI tell us about legal
reasoning in general? Human-computer
interaction (Feb 3, 14–16)
5. Legal technology now: information retrieval,
electronic discovery, knowledge management (Feb
10, 14–16)
11. Lesson plan (2/2)
6. Legal technology now: case management, online dispute
resolution, access to justice (Feb 17, 14–16)
7. Legal technology now: decision support, prediction,
automation, self-service (Feb 17, 16–18)
8. Ethical and regulatory questions. AI and IP law. Big data
and data protection (Feb 24, 16–18)
9. Legal technology in the future: emerging technologies,
innovation, disruption and legal startups (Mar 3, 14–16)
10. Legal technology and you: the impact of legal technology
on the legal profession, new business models for legal
services and alternative business structures, unauthorized
practice versus liberalization (Mar 10, 14–16)
12. Lecture etiquette
- please interrupt me!
- but say your name at least once a day when
you do
- use electronic devices if you absolutely have
to (for taking notes, looking up relevant stuff
etc)
13. Required reading
- Susskind book chapters and/or articles given
on Moodle for each lecture
- not going to go all Socratic on you...
- ... but reading the indicated things at least
cursorily before each lecture should make it
a lot easier to understand the lectures
- and of course you’ll need them for the
paper, supplemental readings also indicated
14. Final paper
- 2500–4000 words (10–16 pages)
- to be returned on Moodle by Apr 10
- topic and form must be approved by the lecturer in
advance
- possible topics (non-exhaustive list):
- some specific technology and its application to law
- some specific field of law or type/stage of legal
practice and the current/potential application of
technology in it
- thorough analysis of 1–2 existing legal startup(s)
- business plan for your own future legal startup
15. Communications protocols
- in person (somewhere here for about 1 h
before each lecture (except 14:00–14:15))
- Moodle
- Twitter: @ronkaine #legaltechturku
- if you absolutely must: email
anna.ronkainen@onomatics.com
- my blog: http://www.legalfuturology.com/
(posts tagged ltcourse)
17. What is technology?
- τέχνη ‘art, skill, craft’ + -λογία ‘study of’
- “Technology is society made durable” (Bruno
Latour)
- ”technologies of power” (Michel Foucault)
- the practical application of knowledge to a
particular area
- “the collection of tools, including machinery,
modifications, arrangements and procedures
used by humans” (yay Wikipedia!)
20. What is legal technology?
- technology (mainly ICT) used
- in courts
- in legal practice
- for doing things which conventionally have
required the assistance of a lawyer
- ...
21.
22. What kind of legal technology does
this course (not) cover?
- ICT only: no photocopiers, no writing
- law-specific only: no e-mail
- focus on innovative technologies: no (or very
little) Finlex or LexisNexis
23. What kind of legal technology does
this course cover?
- artificial intelligence
- machine learning
- cloud computing
- big data
- disruptive
- innovative
- robot judges
- (bingo!)
24. No, seriously. What types of legal
technology does this course cover?
Lecture 5:
- information retrieval
- e-discovery (e-disclosure)
- knowledge management
Lecture 6:
- case management
- online dispute resolution
- access to justice solutions
Lecture 7:
- decision support
- prediction
- automation
- self-service
25. And the other 6.29 lectures?
- brief history of the AI and law field
- emerging technologies most likely to make
an impact
- my own research (sorry, couldn’t resist...)
- professional ethics and regulatory issues
- innovation! disruption!! startups!!!
- ...and what you can/should do about all this
27. Susskind (ch. 3):
The evolution of legal service
1. bespoke
2. standardized
3. systematized
4. packaged
5. commoditized
28. 1. Bespoke lawyering
- the traditional model: everything done
individually for each client
- not going to disappear, high-profile litigation
will certainly always have a lot of this
- however, its role is diminishing
- hourly billing offers no incentives for greater
efficiency to the service provider...
29. 2. Standardized lawyering
- ... but who wants to pay for each contract to
be written from scratch (heck, who even
wants to actually do that)
- standard document templates
- checklists
- the bulk of work still done manually
30. 3. Systematized lawyering
- same as standardized, only with better tech
- e.g. computerized checklists or process
manuals for compliance (workflow systems)
- automated document generation, with a
decision tree logic to select the right type of
document, using just the necessary inputs
31. 4. Packaged lawyering
- systematized lawyering offered so the clients
can use it themselves
- tools and information offered online in
ready-made chunks, backed by individual
(manual) service
- pricing model innovation by this stage, e.g.
based on fees for specific transactions or
monthly/annual subscription fees
32. 5. Commoditized lawyering
- packaged lawyering minus people, and with
even better tech
- offered strictly as a computerized service e.g.
as a web or mobile app
- scalable (the same number of people can
provide the service to 1 or 100000 people),
can be provided at a radically lower cost
- this is what many (but far from all) legal
startups are doing
33. ...and that’s why we’re here
- the role of tech grows at each stage and its
importance for legal innovation is
unquestionable
- but it’s not everything
- an ounce of prevention is worth a pound of
cure!
- design thinking now emerging in law
- alternative dispute resolution
- legal project management
35. Artificial intelligence (AI)
- basically: what people can and computers can’t
(yet) do
- when it becomes possible, it generally starts to
be called something else
- (no good definition for intelligence of the non-
artificial kind in psychology either, except that
it’s whatever IQ tests measure)
- deep AI: general purpose intelligence (cf. the
Terminator movies)
- shallow AI: task-specific intelligence (this is
where the action is for us in law)
36. Turing test
- the most (but not very) agreed-upon validation
experiment for deep AI
- a number of people have to carry a
conversation with a person and a computer
without knowing (or the setup revealing) which
is which and >30% have to get it wrong
- “passed” by “Eugene Goostman” in 2014 (by
lowering expectations by claiming to be a 13-
y.o. non-native speaker of English)
- The Imitation Game: in cinemas Feb 20
37. (the technological)
Singularity
- the moment when the computing power of
all computers combines exceed that of
humankind
- depending on who you believe, the
beginning of the total annihilation of
humanity or total eternal bliss
- (if you ask me, I think the whole issue is not
well-formed)
38. Moore’s Law
- the observation that transistor density in an
integrated circuit doubles every ~2 years
(Gordon E. Moore, co-founder of Intel, in
1965)
- has enabled the exponen-
tial growth of computer
processing capacity
- likely to slow down soon
39. Tests for shallow AI
- games, e.g.
- tic-tac-toe
- checkers
- chess
- Jeopardy!
- poker (2015!)
- go
- football (Robo-Cup)
http://xkcd.com/1002/
40.
41.
42. Rule-based artificial intelligence
- also known as Good Old-Fashioned Artificial
Intelligence (GOFAI)
- based on symbolic (human-readable)
representations of rules, logic etc.
- expert systems
- dominant paradigm in AI through the 1980s
- still the dominant paradigm in AI & law, but
it too is finally starting to give way to...
44. Statistical artificial intelligence
Machine learning
- algorithms created by learning a model of the
target domain from data typically using some
general-purpose algorithm
- supervised learning
- unsupervised learning
- reinforcement learning
- the dominant paradigm in most areas of AI for a
couple of decades now
- enabled by advances in processing capacity and
better availability of teaching data (Big Data)
45. Which methods are the best for law?
- the best (in my opinion: the only) way forward
is to use both where they are best
- rule-base methods are easy to implement and
maintain
- statistical methods can better accommodate
real-world complexity
- vast majority of AI & law research done in rule-
based frameworks (or pure theory), statistical
methods quickly emerging
- (more about this in Lecture 4)