ARTIFICIAL

INTELLIGENCE AND THE

LAW
DR. BUKOLA O. OCHEI
DEPARTMENT OF PUBLIC LAW
 Learning Objectives

1. Understand Artificial Intelligence (AI), have a basic definition of AI and know the major branches of
AI
2. Understand Machine Learning (ML) and the components of ML, appreciate the advantages and
Disadvantages of ML and understand the limits of AI
3. Have an overview of the History of AI and the Law and discuss AI within law and legal studies
4. Have an appreciation of the Benefits and Shortcommings of AI within the Law
5. Consider current and policy issues on AI
 Artificial Intelligence Generally
1. What is AI?
 Rules, Logic and Knowledge Representation
Approach
 Machine Learning Approach
 Hybrid Approach
2. Artificial Intelligence and the Law
 History of Artificial Intelligence and the Law
 Artificial Intelligence in Law Today
3. Current Policy Issues in Artificial Intelligence and the
Law
 What is Artificial Intelligence

 AI is the theory and development of computer systems able to perform tasks normally requiring human
intelligence.
 Artificial intelligence is using computers to Solve Problems or make automated decisions for tasks that
when humans do them typically require intelligence.
 Such as playing chess, driving cars or translating languages. These are all tasks that are thought to have
higher order cognitive processes when people do them
 If a computer can solve problems that are considered as requiring cognitive processes, it is regarded as
an AI task.
 However, it must be understood that computers solve tasks differently from humans as AI tasks involve
the automation of activities that involve human intelligence.
 We live in the age of the data-driven algorithm
 Decisions
that affect our lives are being made by
mathematical models
 Limits of Artificial Intelligence

We don’t have Strong AI

 Computers that think at a level that surpasses humans
 Computers that engage in Abstract thinking
 i.e. thinking that Artificial Intelligence can discuss points of law and facts

There is no Strong AI today – strong AI would be computers that can think at a level that meets or
surpasses people or computers that can engage in abstract reasoning and thinking, conversation in arbitrary
topics. Today’s most advanced Algorithms cannot think nor replicate human higher order of cognitive
reasoning as a two year old has more advanced cognitive abilities than most advanced AI today
 Artificial Intelligence
 AI is predominantly pattern-based AI; this involves the use of algorithms to search large amounts of data or patterns
that can be channeled and used. The formal name for this is, Machine Learning.
 Pattern-based AI is a very formidable tool that has been used to automate many processes today; such as language
translation, fraud detection and driving.
 In limited domains, the current AI technology can do wonderful things, it is however not unlimited in what it can do.
 It also includes the ability of computers to adapt to new circumstances and is regarded as a combination of different
abilities
 Although pattern based AI is the dominant mode of AI currently, there is also the rules-based logical AI systems
which is used to a lesser extent.
What Artificial Intelligence Currently Is

 AI is the theory and development of computer systems able to

perform tasks normally requiring human intelligence, such as:
 Visual perception
 Speech recognition
 Decision-making
 Language translation
 Major Branches and Approaches to AI

AI can be roughly divided into two large groups

i. Computer Logic and Rules-Based (XR) Approach
ii. Machine Learning (Pattern-Based) Approach
 Logic, Rules and Knowledge Representation Based
AI

 Rules and knowledge based systems are part of a larger category of Artificial
Intelligence also known as knowledge representation.
 The aim of knowledge representation is to model real world processes or systems in
ways that a computer can use and process.
 This means that programmers or groups of engineers observe real world tasks, often
involving experts, which has certain real world constraints and rules.
 Then they attempt to model that real word process in computer form in terms of
computer rules and relationships
A good example of this AI within law- is the ‘Turbotax expert system software’ for tax assistance which
people use to assist them in tax related issues. For instance, the ‘Turbotax’ software can compute tax
liability in terms of Personal Income Tax Laws. The ‘Turbotax’ software is an example of a knowledge
representation based version of AI.
The company that created Turbotax endeavored to model something out there- personal income tax law, a
set of real world constraints on people’s legal liabilities and duties – they modeled peoples’ personal
income tax law as a software in a way that it represents the logic and the meaning in the laws.
The programmers, in consultation with lawyers and accountants studied the personal income tax laws and
reduced it to a set of rules and attempted to translate the rules into a set of comprehensible rules that the
computer could understand. This is known as an expert system where knowledge of experts such as
lawyers and accountants has been reproduced computationally and patterned to produce real world results
such as tax liability
 What Machine Learning Is

 Algorithms that find patterns in data and infer rules on

their own
 Learn from data and improve over time
 Use the patterns to automate or predict
In this context, it is best to consider ML as a group of things
as opposed to a single thing
Uses of Machine Learning

 Self-driving cars
 Netflix, Showmax – automated recommendations
 Google translate – computer translation
 Characteristics of Machine Learning

1. Learning – algorithms are specifically designed to “learn”. They are learning because they improve on their
performance over time, often by being shown good and bad examples of what to do.
AI can apply different forms of learning such as learning by trial and error, memorizing individual items and procedures, known
as rote learning, also, generalization; applying past experience to analogous new situations
2. Pattern Detection - Much of machine learning can be thought of as pattern detection systems designed to detect
non-obvious patterns of behavior or correlations in large amounts of data. These patterns are then generalized and used
in other contexts to make useful decisions.
3. Data -Machine learning requires data, often, large amounts of data to detect the patterns that can be used to
automate decisions. In law, this can be a problem because not a lot of legal information is available in highly
processible form, much of it is behind paywalls or non-accessible structures.
4. Self-Programming- ML is regularly considered as self-programming, because self-programming agents record data
in memory as they learn and this program can control the execution of learned data as sequences of instructions
Key Note on ML

 Intelligent Results without intelligence- ML in general does not have the cognitive abilities of humans, it
therefore achieves the “intelligence” requirement through proxies or stand-ins.
 When proxies and stand-ins can be found, the computer does not have to understand the deep conceptual
layer in order to make a decision.
 AI does not know the meaning of words like “free” or “spam”
 It gets “intelligent” automated results without intelligence by finding suitable proxies and patterns to make
useful automated decisions without actually having underlying higher orders of human intelligence.
The Proxy Principle for Automation
 It has to do with detecting patterns that can serve as proxies for some underlying cognitive task.

Hybrid Systems
Although there are two major AI techniques –Logic and Rules Based Approach, and Machine Learning(Pattern Based
Approach), may successful AI systems are hybrids of ML and Rules-Based Hybrids.
An example of this is self-driving cars which employ both approaches effectively.

Human Intelligence + AI Hybrids

There is a misunderstanding that AI is replacing humans, but that is not true, rather, it is most times better to think about
it as enhancing humans. Also, many successful AI systems work best when;
 They work with human intelligence
 AI systems supply information for humans
 Many AI systems have humans in the loop
 Human Intelligence + AI Hybrids Contd.

 With a combination of human intelligence and AI Hybrids, humans are able to make better
decisions. For instance, although a computer may be the best computer player in the world,
a combination of Human Intelligence and AI can beat many of the best computer programs
because the human chess players abilities are being enhanced by the computers.
 Although AI is often thought of as being autonomous, many AI systems still have humans
in the loop. The idea is that a computer will often ask for help from a human to make a
judgment decision or an abstract conceptualization about an area where the computer is not
able to perform.
As such, AI systems are not fully autonomous.
 Ways AI Will Benefit the Law
 AI can be helpful; it can enable improvement in products and services, not replace humans that use them.
It can improve productivity and efficiency through the use of software that can draft documents, improve
search and extraction of relevant content for analysis – KIRA, COINC, RAVEL
 It can save time, money and improve efficiency by managing emails, billings and productivity.
 In the context of legal research and analytics, AI is capable of sifting through millions of legal precedents
to find relevant case law in seconds, it can also review judges sentiments in any issue of law and also
analyse opposing counsel’s style and reasoning on specific issues, based on his past representations.
According to Aruda’s TED Talk, ROSS (a software developed by a Nigerian) can read over a million pages
of law documents in seconds, find the exact passages needed. This is an example of how AI levels the
playing field when it comes to research.
It makes you proactive as opposed to defensive, it connects the dots and gives an advantage.
 History of Artificial Intelligence and Law
 Early thinkers
Gottfried Wilhelm Leibniz (1600’s) – one of the co-inventors of calculus, was both a mathematician and a
lawyer. As far back as the 17th century, Leibniz postulated that law could be made more predictable, more
measurable, more useful. This is the heart of AI in law, using computational and mathematical formula to
make law more understandable or accessible, predictable, manageable and useful.
Leibniz contributed to mathematics, as with many lawyers who have made important contributions to
mathematics. For instance, the mathematical concept of Linear algebra which is basically the mathematical
backbone of most of machine learning currently, was founded by lawyers; Arthur Cayley and James
Sylvester were both lawyers and mathematicians.
There is a long tradition of lawyers influencing mathematics in the interconnectivity between law and
technology
Pre-AI Modern Era
Lee Loevinger (Jurimetrics)- The real modern era of AI in law started prior to the emergence of computing and can be said
to go back to Loevinger whose idea of jurimetrics evolved in the 1940’s
Era of Knowledge Representation
The heart of AI and Law started in the 1960-1980’s where we can see the early pioneers of AI and Law such as Ashley K;
McCarthy, T. (coined the term ‘Artificial Intelligence); Gardner, A.; Hefner C.; Rissland , E.; Sussland, A.; Allen, L.;
Sergot, M. Kowalski, R.; Winkels, R.; Sussland, R.; Governatori, G.; Shannon C.
Many of the early pioneers of AI, who were computer scientists started the idea of representing rules to law. Here, there
were attempts to model legal arguments formally in ways that computers could understand. They endeavored to model
legislation and regulations in ways that computers could understand, through legal text extraction and statutory modelling.
Some important ideas came out of the pre-modern era, such as JURIX and IAAIL which still exist today. They are
conferences dedicated to AI and Law.
Most of the AI in research occurred in Europe, Italy, Netherlands, England and Germany. They were the hotbeds of AI and
continue to be so, even till now.
Present Era – 2000 till date, the Legal Tech and Machine Learning Era
Innovations in AI continued into the 21st century and led to the founding the following
Codex
Legal Tech Startups such as those in Nigeria like LawPadi, Lawyard and on the
international arena, like, Ravel Law, Luminance, Enoron, Farewell, and Lexoo
Legal Analytics and predictions (eg. Lex Machina)
Amazon
Pandora
Google
Tesla
 Three Categories of AI in Law Today
There are three categories of AI in law currently:
1. Administrators of Law – these are people who create and apply the law; Judges, Legislators, government officials and
regulators
2. Practitioners of Law – legal practitioners and others interpreting or advocating in the legal system. Here we see e-
discovery and technology assisted review, legal analytics, venue analysis and case outcome predictions come into
play. Also, legal document assembly, automated document analysis and diligence, natural language processing (NLP)
of legal documents, computable contracts, legal research, law firm management, case management, docketing and
work flow systems
3. Users of Law - ordinary individuals or businesses that have to comply with the law, such as;
i. Businesses which have to comply with regulation such as the General Data Protection Regulation (GDPR and the
Nigerian Data Protection Regulation (NDPR 2019). They also utilize computable contracts, expert systems,
automated dispute resolution and automated legal documents.
ii. There are also ordinary people who utilize legal expert systems , legal chatbots and automated legal self-help,
automated legal document assembly, amongst others
 Shortcomings of AI/What AI is Bad AT
 It raises ethical questions as deep learning algorithms which are the basis of many of the most
advanced AI tools are only as smart as the data they are given in training.
 There is a potential for human bias which must be monitored closely
 There is always a need for the indispensable human element in AI
 Also, AI cannot replace the following – Judicial Roles, Client engagement, development in
areas of institutional voids ( AI cannot develop new business areas such as medical law, sports
law, construction law etc), and government relations, such as developing legislation and
regulations.
These require an understanding of people, government and even the political landscape. AI
cannot do these.
 Shortcommings Contd.
 AI cannot perform abstract, conceptual tasks , as such, lawyerly tasks involving are less susceptible to being swept
away by AI.
i. Abstract thinking
ii. Problem solving/ Advocacy
iii. Client counselling
iv. Human emotional intelligence
v. Policy analysis
vi. Creative thinking
vii. Big picture strategy.
All these tasks are less likely to be subject to AI’s automation in the near future, given the limit of the current AI
technology. As such, these are the areas that intending lawyers should focus on.
 Ways AI Will Benefit the Law
 AI can be helpful; it can enable improvement in products and services, not replace humans that use them.
It can improve productivity and efficiency through the use of software that can draft documents, improve
search and extraction of relevant content for analysis – KIRA, COINC, RAVEL
 It can save time, money and improve efficiency by managing emails, billings and productivity.
 In the context of legal research and analytics, AI is capable of sifting through millions of legal precedents
to find relevant case law in seconds, it can also review judges sentiments in any issue of law and also
analyse opposing counsel’s style and reasoning on specific issues, based on his past representations.
According to Aruda’s TED Talk, ROSS (a software developed by a NIgerian) can read over a million pages
of law in seconds, find the exact passages needed. This is an example of how AI levels the playing field when
it comes to research.
It makes you proactive as opposed to defensive, it connects the dots and gives an advantage.
 Current Policy Topics with AI and the Law
These are what AI researchers and others are considering about AI today
Automation of Legal Jobs
Some tasks in law that were traditionally done by lawyers are being automated by AI. Lawyers are nowdoing a mixture
of tasks that run from highly abstract to the routine and mechanical. The mechanical legal tasks will likely be
automated once there is an underlying structure or pattern that can be harnessed.
Creation of New Legal Jobs
Just as technology is taking away, it will also give and this has been a constant in the evolution of man. For instance,
human calculators were replaced by computers in the 1950’s.
In the short term, some jobs will be lost in law due to AI, but new jobs also emerged, created by AI; computer game
designer, web programmer, systems analysts etc.
It becomes necessary to think about the social impact of AI not just to law but to the Nigerian society generally
 Other Areas in AI

 Natural language processing - Natural language processing (NLP) is a branch of

artificial intelligence that helps computers understand, interpret and manipulate
human language. NLP draws from many disciplines, including computer science and
computational linguistics, in its pursuit to fill the gap between human communication
and computer understanding.
 Fuzzy logic – Fuzzy logic is a method of machine learning designed to view the
world in a more human manner and gain additional knowledge. A prime example is a
chatbox learning the natural language of humans
 Other Areas in AI contd.

 Neural Networks- Neural networks are systems that try to emulate the human
brain, combining computer science and statistics to solve common problems
in the field of artificial intelligence. They are composed of layers of nodes, or
artificial neurons, that can process and transmit information. Neural networks
can learn from data and improve their performance over time. They are used
for various tasks such as speech recognition, image recognition, natural
language processing, amongst others
 Other Areas in AI contd.

 Internet of things – or IoT, refers to the collective network of connected devices and the technology that facilitates
communication between devices and the cloud, as well as between the devices themselves. This is in a large part,
facilitated by the advent of inexpensive computer chips and high bandwidth telecommunication, that allows billions
of devices to be connected to the internet. This means everyday devices like toothbrushes, vacuums, cars, and
machines can use sensors to collect data and respond intelligently to users.
The Internet of Things integrates everyday “things” with the internet. Computer Engineers have been adding sensors and processors
to everyday objects since the 90s. However, progress was initially slow because the chips were big and bulky. Low power computer
chips called RFID tags were first used to track expensive equipment. As computing devices shrank in size, these chips also became
smaller, faster, and smarter over time.
All “invisible computing devices” and the technology associated with them are collectively referred to as the Internet of Things.
 Other Areas in AI Contd.

 Deep Learning - Deep learning is a machine learning technique that teaches computers to do
what comes naturally to humans: learn by example. Deep learning is a key technology behind
driverless cars, enabling them to recognize a stop sign, or to distinguish a pedestrian from a
lamppost. It is the key to voice control in consumer devices like phones, tablets, TVs, and
hands-free speakers. Deep learning is getting lots of attention lately and for good reason. It’s
achieving results that were not possible before.
 In deep learning, a computer model learns to perform classification tasks directly from images,
text, or sound. Deep learning models can achieve state-of-the-art accuracy, sometimes
exceeding human-level performance. Models are trained by using a large set of labeled data
and neural network architectures that contain many layers.
 Other areas in AI Contd.

 Big data- Deep learning is a machine learning technique that teaches computers to do what comes naturally to humans: learn by example. Deep learning is a key
technology behind driverless cars, enabling them to recognize a stop sign, or to distinguish a pedestrian from a lamppost. It is the key to voice control in consumer
devices like phones, tablets, TVs, and hands-free speakers. Deep learning is getting lots of attention lately and for good reason. It’s achieving results that were not
possible before.
In deep learning, a computer model learns to perform classification tasks directly from images, text, or sound. Deep learning models can achieve state-of-the-art
accuracy, sometimes exceeding human-level performance. Models are trained by using a large set of labeled data and neural network architectures that contain many
layers.
Systems that process and store big data have become a common component of data management architectures in organizations, combined with tools that support big
data analyisis uses. Big data is often characterized by the three V's:
• the large volume of data in many environments;
• the wide variety of data types frequently stored in big data systems; and
• the velocity at which much of the data is generated, collected and processed.
These characteristics were first identified in 2001 by Doug Laney, then an analyst at consulting firm Meta Group Inc.; Gartner further popularized them after it acquired
Meta Group in 2005. Recently, several other V’s have been added to different descriptions of big data, including veracity, value and variability.
 Although big data doesn't equate to any specific volume of data,big data development often involve terabytes, petabytes and even exabytes of data created and
collected over time.
 Other areas in AI to consider contd.

Block Chain - a block chain is a digital database containing information

(such as records of financial transactions) that can be simultaneously used
and shared within a large decentralized, publicly accessible
networkalso : the technology used to create such a databaseThe technology
at the heart of bitcoin and other virtual currencies, blockchain is an open,
distributed ledger that can record transactions between two parties
efficiently and in a verifiable and permanent way.