4. What is AI?
Definition (Merriam-Webster)
1: a branch of computer science dealing with the simulation of intelligent
behavior in computers
2: the capability of a machine to imitate intelligent human behavior
9. Exercise 1
What terms should you include in a
contract to protect the parties’ interests?
10. You are buying a car: what promises do you
demand from the seller?
You are leasing a car: what promises do you
demand from the lessee?
Title
Specifications
Warranty
Insurance
Maintenance
Milage and overage
Return
15. Simple Example: an app to choose a car
Step 1: what cars are available?
Profile: catalog all information about cars
Find relevant material: searchable database about cars
Step 2: what are the characteristics of each car?
Identify relevant characteristics: (e.g. price, capacity, mileage, type)
Step 3: what is the best car for me?
Predict the best car: map car characteristics to buyers
17. Act II: Search (find relevant material)
Manual
Human selection of search terms
Automated Training
Machine analysis of exemplars and
training sets
Core Technology
● predictive coding
● term frequency-inverse
document frequency (“tf-idf”)*
*83% of text-based recommender systems in
the domain of digital libraries use tf-idf
18. Act III: Analyze (identify relevant characteristics)
Manual
Human categorization (e.g., Dewey
Decimal Classification System)
Automated Deconstruction
Machine identification e.g., issues
in a case; elements of a deal
Core Technologies
● Structural Classification
○ Deconstruction
○ TF-ICF (at clause level)
● Text similarity measures
○ Levenshtein
○ Cosine
● Natural language processing
19. Act IV: Optimize (predict best outcome)
Manual
Ask the expert
Automated Expertise Systems
Machine identification of the
characteristics yielding optimal
results
Core Technologies
● Program languages; Neota Logic
● Pattern matching (Jaccard)
● n-Grams; skip-grams
● Neural networks
● Word embedding (Word2vec)
● Deep learning
23. Clause 1 Clause 2
Each party (the "Indemnifying Party") shall indemnify, defend and hold
the other party, including, in the case of Company, Company's landlord,
and each party's affiliates, directors, shareholders, officers, agents,
third-party vendors, licensors, lenders, independent contractors,
employees, successors and assigns (each, an "Indemnified Party" or
collectively, "Indemnified Parties") harmless from and against any and
all claims, damages, costs, liabilities, losses and expenses (including
reasonable attorneys' fees and out-of-pocket expenses) ("Losses")
resulting from any claim, suit, action, demand, or proceeding (each, an
"Action") brought by any third-party against the Indemnified Party (i)
alleging or arising from the gross negligence or willful misconduct of the
Indemnifying Party or its employees, agents, contractors, or invitees, in
the performance or non-performance of its obligations hereunder or
upon access to the Data Center, (ii) arising from any failure by the
Indemnifying Party or its employees, agents, contractors, or invitees, to
comply with the applicable law, and (iii) arising from or connected with
any claims of ownership or superior rights to Customer Equipment or
Company Equipment or other intellectual property by a third-party.
Customer will indemnify, hold harmless, and defend Company, its
officers, directors, parent and/or affiliated companies, employees,
agents and subcontractors from liabilities, claims or damages and
expenses whatsoever (including reasonable attorney's fees) arising out
of or in connection with Customer's negligence and/or damages to
Company's tangible personal property. Customer's indemnification
obligations do not apply to claims for damages to real or tangible
personal property or for bodily injury or death which is solely caused by
Company due to Company's gross negligence or willful misconduct.
How to identify the patterns (the concepts)
24. Clause 1 Clause 2
Each party (the "Indemnifying Party") shall indemnify, defend and hold
the other party, including, in the case of Company, Company's landlord,
and each party's affiliates, directors, shareholders, officers, agents,
third-party vendors, licensors, lenders, independent contractors,
employees, successors and assigns (each, an "Indemnified Party" or
collectively, "Indemnified Parties") harmless from and against any and
all claims, damages, costs, liabilities, losses and expenses (including
reasonable attorneys' fees and out-of-pocket expenses) ("Losses")
resulting from any claim, suit, action, demand, or proceeding (each, an
"Action") brought by any third-party against the Indemnified Party (i)
alleging or arising from the gross negligence or willful misconduct of the
Indemnifying Party or its employees, agents, contractors, or invitees, in
the performance or non-performance of its obligations hereunder or
upon access to the Data Center, (ii) arising from any failure by the
Indemnifying Party or its employees, agents, contractors, or invitees, to
comply with the applicable law, and (iii) arising from or connected with
any claims of ownership or superior rights to Customer Equipment or
Company Equipment or other intellectual property by a third-party.
Customer will indemnify, hold harmless, and defend Company, its
officers, directors, parent and/or affiliated companies, employees,
agents and subcontractors from liabilities, claims or damages and
expenses whatsoever (including reasonable attorney's fees) arising out
of or in connection with Customer's negligence and/or damages to
Company's tangible personal property. Customer's indemnification
obligations do not apply to claims for damages to real or tangible
personal property or for bodily injury or death which is solely caused by
Company due to Company's gross negligence or willful misconduct.
Who indemnifies who, for what, and under what circumstances?
26. Limitations and Observations
Gary Marcus
“We wanted Rosie the robot, and instead we got the Roomba.”
His point
Current AI is not a generalized artificial intelligence; however, deep
learning can be successfully applied to specific objectives
Observation: AI is a inductive reasoning process...it, like deductive
reasoning, has been around centuries…so what’s all the hype about?
27. uncommon + divergent
Deal-Specific
AI is a tool that offers insights that can be trained in minutes, not years
Negotiated
common + divergent
Master Services Agreement (IT)
commonality
consistency
Standard
common + consistent
uncommon + consistent
Transaction-Specific
29. Application to Contract Analysis
Determines
● how contracts are organized
● what clauses they contain
● standard language for each
clause
● full range of deal-specific
alternatives
Metadata ExtractionStructural Classification
Captures
● presence or absence of
clauses
● key contract language
● key business terms
○ names
○ places
○ dates
○ amounts
source contracts
Identifies
● subject
● verb (actions)
○ what must be done
○ what can’t be done
○ what can be done
● object
Natural Language
30. Evolving Technology
What do you want to find?
How is it organized?
How will you find it?
Manual +
Discrete blocks
Manual
Code
Manual +
Discrete blocks
Semi-Automated
Machine Learning
Automated +
Hierarchical facets
Automated
Deep Learning