This presentation was given at Legal Geek on 10th Dec 2015. It is a scenesetting peice that looks to de-mystify artificial intelligence by looking beyond the hype.
What AI is and examples of how it is used in legal
1. What AI is and examples of how it is used in legal
Ben Gardner
Information Architect
Legal Geek – December 2015
Email: Ben.Gardner@Linklaters.com
LinkedIn: https://www.linkedin.com/in/bengardner135
2. Why are we talking about AI?
2015 Law Firms in Transition - An Altman Weil Flash Survey
“Can you envision a law-focused ‘Watson’ replacing any of the following
timekeepers in your firm in the next 5 to 10 years?”
3. Overview
1. What Artificial Intelligence is and is not
2. Artificial Intelligence in Legal
3. What does this mean for legal?
6. Narrow or Weak Artificial Intelligence
Weak AI
Statistical Symbolic
Requires
scale
Hand
crafted
Translation services
Google Now
Siri/Cortana
Watson
Wolfram|alpha
Models
Natural Language Processing
Machine Learning
“Weak AI - non-sentient computer intelligence or AI that focuses on one narrow task” - Wikipedia
7. It’s all about the models
Statistical
Regression
• Predict a real-value response (viral
load, weight)
• Gaussian, Gamma, Poisson and
Tweedie
• MSE and R^2
Classification
• Multi-class or Binary classification
• Ranking
• Accuracy and AUC
Clustering
• Unsupervised learning (no training
labels)
• Partition the data/identify clusters
• AIC and BIC
Symbolic
Ontology
16. Evolution of digital information management on the web
Yahoo & Altavista Google Synonyms Knowledge Graph ???????
Categories &
Keywords
PageRank
Semantic
Machine
Learning
Ontologies
Linked Data
Natural Language
Processing
Reasoning
Taxonomies
‘98 ‘03 ‘12‘94 Hummingbird‘13
Web 1.0 Web 2.0 Web 3.0 Web 4.0
Ben Gardner
Vocabularies
17. Artificial Intelligence in Legal
1. Text analytics
2. Expert systems
3. Ross the super intelligent attorney
4. Start-ups
18. Text analytics
Vocabularies Synonym matching
Entity Extraction Identify “things” rather than “strings”
Natural Language Programing Analysis of the contextual meaning of
text e.g. questions
Machine Learning Document clustering
Identification of clauses
19. Applying Text Analytics
Due Diligence Document Comparison
• Capacity to compare
documents against a
standard template and red
line across corpus
• Identify deviations form the
norm and apply analysis
• Automatically identify
clusters of documents within
corpus of documents
• Automate the extraction of
data from document content
• Capacity to analysis large
volumes of documents
rather than a subset
20. Expert Systems
• Classical approach to AI
• Powerful approach to
guiding users through
complex decision making
• Automating triage of
structure content
26. Innovating Services & Delivering Efficiency
Scenario
1. Perform Due Diligence over a corpus of 100000 contracts
2. Extract contextual information i.e. Borrower, Interest rate, Change of
control clause, etc.
3. Link Borrower to Thompson Reuters companies data to enrich with
profile information
4. Use expert systems to triage/categorise based on extracted & enriched
information and automatically identify action
What else could you do with this?
Text
analytics
Enrich
information
Expert
system
27. Final thoughts
Is Artificial Intelligence a threat or an opportunity?
• AI is an agent for change
• Is the demand for legal product elastic or inelastic?
• Will this be a Red Queen Race?
Bespoke Standardised Systematized Packaged Commoditized
Innovation
Evolution of legal services - Richard Susskind
Continuous Improvement
28. Nothing new!
“We recommend that all solicitors
should review their practices and
office procedures to ensure that
they take full advantage of the
efficiency and economy which can
be obtained with the assistance of
modern technology.”
Royal Commission on Legal Services
(aka the Benson Commission) - 1980
Legal Insider - http://www.legaltechnology.com/latest-news/throwback-thursday-first-thing-we-do-lets-computerise-all-the-lawyers-from-1980/