AI and the Law (Part II) - How AI Works

  • Posted on: 31 March 2017
  • By: warren

(Note: This is the second part of a series of posts that were based on several conversations with lawyers and executives about AI, the nature of technology and its application to business problems. The first part is here.)

At the heart of it, AI is about asking the following question: "Can I use the computer to make decisions that would normally require a human being?" Of course, the obvious answer is yes; human beings make all sorts of decisions all day long, ranging from the complex to the mundane. Accounting and operations systems have been making decisions for human beings for years, be it from calculating credit scores and interest rates to determining the best time to order feedstock.

Let's use a simple example to explain the difference. Take the plot on the right hand side where I have orange dots and green dots. With basic statistical methods (linear regression was invented in the early 1900's[1]), we can create a simple classifier that will separate the green from the orange by simply drawing a line through the graph. It's not perfect, some oranges are misclassified as green and vice-versa, but we do very well with a really simple method. We can do better using mathematical techniques that are more sophisticated, see a non-linear method on the left, but fundamentally the problem remains simple: telling the granny smith apples from the oranges.

The problem is simple to solve because it is well defined. The objective is clear, the definition of success is clear (keep both sets of dots separated) and the way to tell them apart is by their colour. The only thing that remains is applying the recipe that matches the problem, a linear regression in this case, to solve the problem. Loosely speaking, there is no intelligence needed because the problem defines its own process to a solution.

Similarly, let's take another toy problem: tic-tac-toe. Every school child, even those that don't eventually end up working on AI, learns that the game can be tied or won by the first person playing. The second player can always force a tie, but can never win the game. All of us learn this by playing the game repeatedly while young: children over time explore the set of possible game layouts in tic tac toe and eventually learn that there are a finite number of starting X's and O's that can lead to victory or loss.

Very roughly, there are over 360,000 possible positions in tic-tac-toe. One basic machine learning method to learn to play tic-tac-toe is brute-force: try every single move and counter-move until every game is enumerated and then only choose moves that end in winning the game. Obviously, children can't keep track of 360,000 tic-tac-toe boards at the same time and while effective, the method does not scale well even for computers (Your desktop computer can't store the 10^47 possible combinations of chess). Therefore, children learn to take shortcuts and reduce those 360,000 positions into a set of starting moves that ensures that they win the game every time. That is intelligence: no one taught them the process required to find the solution, they just did. Artificial Intelligence is the science of creating algorithms that can do the same for certain classes of problems.

We say "classes of problems" because we only have the computing power and the programming know-how to handle limited problems, like recognizing a person, a text or a musical genre. This is different from what people sometimes refer to as "True AI", which is the human-like machine that walks and talks (and invariably tries to take over the world) on television shows. In my opinion, you are unlikely to have the Terminator do your filing for you anytime in the near future.

However for specific types of problems, AI works very well: classification, clustering, searching, reduction, etc... In turn, this means that the most of the work that goes into implementing an AI engine is actually trying to match very simple mathematical solutions to complex business problem. Going back to our first example of apples and oranges, the problem was delivered on a plate: colour and position. The solution needs more thinking when the problem isn't so well defined. In some cases, we know that some of the dots are different but not why or which ones (eg: Outlier Detection). In others, the objective may be to "Group the similar dots together" without having any idea of what makes the dots similar (eg: Market Segmentation). 

Machine Learning vs Artificial Intelligence

In the vernacular, the terms Machine Learning and Artificial Intelligence are sometimes used interchangeably through they refer to different things. Artificial Intelligence is the catch all phrase for different computational techniques that have an intelligence component to them irrespective of the flexibility or adaptability of the method. Machine Learning refers to methods that are capable of learning themselves from the data without having their decision model encoded by a human being.

Take a program that can play tic-tac-toe. It clearly has an intelligence component in order to function, but the software will not learn from its interaction with the user (The problem is simple enough that there is no point to). But a program that recognizes cats in videos needs a machine-driven learning component is order for it to learn what a cat looks like.

Classifying And Finding Things With Artificial Intelligence

The figure below represents a very simplified block diagram of the AI process for classification; starting from the left to the right. We have a data set that we want processed, which can be documents, images, songs, video, etc... In practice, not everything within that data set is relevant to the context of what we are trying to do and so we transform each document into a set of features that we think are valuable to solve our problem. A feature might be a specific word in a document, another might be the word's part-of-speech (verb, noun, adverb, etc) or a typographical aspect such as the word being underlined.

Since the feature set explicitly determines what part of the data set the algorithm will actually look at, feature generation is an extremely important part of the Artificial Intelligence process. It has spawned it's own field of study, Feature Engineering, and at times some have insisted that Artificial Intelligence is carefully crafted Feature Engineering. In practice, many engines have enough computational resources that they will simply generate every feature possible from the input data and the algorithm will simply choose the features that are most promising (This is called the "Throwing things at the wall to see what sticks" approach). It's wasteful, but computing time has become much cheaper than the people time required to create an efficient design.

The algorithm is the brain of AI, which is ironic in that the algorithm in itself is usually very simple and generic; the same algorithm that flies a flying drone might be the same that keeps your phone camera images from being blurry. However, the devil is in the details and the implementation of the algorithm is usually not portable from the phone to the drone. Examples of algorithms might be k-means, C4.5 or okapi, each one trying to perform the task of clustering, rule generation or information retrieval. As part of the process, the algorithm will take in the features and select the most promising. As part of that selection, some external information such as a trained model or parameter might be provided to the algorithm to guide it's decision making.

The results are then checked against a benchmark, sometimes called a gold standard, to ensure that the system is doing what it is supposed to. If the results aren't exactly what is required the model or the parameters of the algorithm might be changed.

Overall, the basics of AI aren't that complex, but its implementation and arrangement needs to be focused on the objectives of the project, otherwise one gets into the loop of "garbage in, garbage out". Depending on the case, the tuning of parameters can be frustrating and model generation becomes an art and not a science. There are many different frameworks, libraries and code bases available both freely and commercially to experiment with which I encourage you to do.

Next: Part III - Algorithmic Bias is good for you.

AI and the Law (Part 1)

  • Posted on: 30 January 2017
  • By: warren

(Note: This is the first part of a series of posts that were based on several conversations with lawyers and executives about AI, the nature of technology and its application to business problems.)

What is Artificial Intelligence?

Does it really represent an improvement over what we already have? An entirely new class of solutions to ongoing problems? Or the flavour of the week in a market that is overwhelmed with buzzwords?

Skepticism is endemic to technology culture in industry, government or academia. It's a byproduct of working in an area who's foundation is innovation and ideas. When it costs significantly less to say that you have something than to actually get it to work: a "show me" attitude is necessary. Ironically, IT has been so far primarily about what we would call classical Management Information System. The software may be really slick, the hardware may be really fast and we can store a lot of data, but really most of what the industry has focusing on so far is simply replacing physical forms and paperwork with the electronic equivalent: tabulating ledgers for accounting, generating reports and mailing checks and invoices over the Internet instead of in paper form. These are boring, unglamorous tasks but they have been IT's big success: taking things that were mundane, repetitive and that were cost centres and streamline it using technology.

And now, we have Artificial Intelligence.

Image of a copper engraving from Karl Gottlieb von Windisch's 1783 book Briefe über den Schachspieler des Hrn. von Kempelen, nebst drei Kupferstichen die diese berühmte Maschine vorstellen.

The underlying idea that a machine would replace a human being in making decisions isn't all that new. One of the better know historical exhibits (and fraud) is The Turk, a mechanical automaton that would play chess against a person. Of course, playing chess was asking a lot of simple clockwork mechanisms and the builder had constructed a false compartment in which a human player would hide and move the mannequin using a system of pulleys and cams. This may have been the first vapourware product ever, but the idea that a machine could perform tasks at a human level had taken root.

What we'll call modern Artificial Intelligence appeared in the mid-1950's when scientist began to look at ways that elements of human cognition could be modelled using mathematics. That in itself wasn't novel, humankind had moved on from the abacus. What they were aiming at were higher cognition functions like learning from examples and extrapolating solutions for problems that the machine had never seen before.

In this series of blog posts, the basics of AI will be reviewed and its application to practical business problems outlined. As with many technologies it has had its false starts and the causes of the AI Winters periods will be reviewed which in turn, will give a sense of why it is making such a resurgence.

Next: Part II - How AI Works.

Presentation at Derby University: Artificial Intelligence and the Law

  • Posted on: 21 November 2016
  • By: warren
Artificial Intelligence and the Law
Wednesday, 30 November 2016 at 5.30 PM


This talk is dedicated to the memory of David F. Evans.


After years of over-promising and under-delivering, and two so-called “winters”, Artificial Intelligence is aggressively making inroads in the legal and financial markets. Affordable computational power, bandwidth, and storage each played a part in this revival, but adoption is really the result of targeted products that do what machine learning is historically excellent at: domain specific applications. Concerns over lost employment, reduced professional prestige, and the accountability of such approaches (Do algorithms behave ethically? Should they be regulated?) remain a hot topic for the layman and the practitioner alike as the technology finds new niches to occupy.
This talk is about some lessons learned about the application of AI to legal document review and some of the unexpected corner-cases found along the way, including creative system training methods, tackling entrenched cultural beliefs, and the thorny (and multifaceted) issues of information security.
Working on these problems has highlighted the value of human judgement, the limits of computation, and how scaling problems affect both people and machines. More than ever, sound theory, good mathematics, and a well-rounded approach are needed to tackle an increasingly complex and globalized world.

Presentation to the Security Group, University of Kent

  • Posted on: 15 November 2016
  • By: warren
Processing legal documents with AI: Notes from the field.
Wednesday, 23 November 2016, 5PM
In a world where the marginal cost of copying information is nil, the marginal cost of storing information is nil, and the marginal cost of transmitting information is nil, the opportunities for data disasters are numerous and their consequences are devastating to individuals and corporations. Even well-intentioned gestures such as AOL's release of anonymized query logs have resulted in privacy violations in ways not immediately obvious to the principals.
Interestingly, portability and interoperability are now working against us: in the world of single sign on, once you have access to the user you have access to all of their systems, and the first leak tends to be the last leak. This talk will be about the security of AI engines in contract analysis and the unexpected and counterintuitive lessons learned along the way. Security auditors want audit logs; forensics-aware systems people want traceability; privacy advocates want minimal information kept; developers want ease of maintenance and debugging; and users want to be minimally impacted. Now all of these views are necessarily compatible and some interesting conflicts arise in the creation and operations of applications.
In the end, the creation of online apps that handle sensitive information, preserve privacy and security, while enabling distributed teams to collaborate requires much more than following best practices. They rest on a culture of security within the organization.

Presentation at the Canadian Linked Data Summit: Operationalizing Linked Open Data

  • Posted on: 12 October 2016
  • By: warren
CLDI LogoOperationalizing Linked Open Data
Venue: University of Montreal, 3200 Jean-Brillant, Room B-2245
Monday, October 24th 2016, 14:10 - 14:30
This talk summarizes the combined experiences of the Muninn Project and the Canadian Writing Research Collaboratory in operating large linked open data projects. Topics that will be touched on will include best operating practices, known pitfalls and realizing the promise of the semantic web for researchers.  
Presentation slides in English and in French.

Presentation at Museums on the Web 2015

  • Posted on: 9 March 2015
  • By: warren
Palmer House Hilton, Chicago, IL, USA
April 8-11, 2015, 10:30am - 12:00pm
Grand Ballroom (4F) 
Joint work with David Evans, Minsi Chen, Mark Farrell and Daniel Mayles.
We review the possibilities, pitfalls, and promises of recreating lost heritage sites and historical events using augmented reality and "Big Data" archival databases. We define augmented reality as any means of adding context or content, via audio/visual means, to the current physical space of a visitor to a museum or outdoor site. Examples range from simple prerecorded audio to graphics rendered in real time and displayed using a smartphone.
Previous work has focused on complex multimedia museum guides, whose utility remains to be evaluated as enabling or distracting. We propose the use of a data­-driven approach where the exhibits' augmentation is not static but dynamically generated from the totality of the data known about the location, artifacts, or event. For example, at Bletchley Park, reenacted audio conversations are played within rooms as visitors walk through them. These can be called "virtual contents," as the audio recordings are manufactured. Given that a number of documentary sources, such as meeting minutes, are available concerning the events that occurred within the site, a dynamic computer-generated script could add to the exhibits.
Visitors' experiences can therefore react to their movements, provide a different experience each time, and be factually correct without requiring any expensive redesign. Furthermore, the use of a data-driven approach allows for the updating of exhibits on the fly as researchers create or curate new data sources within the museum. If artifacts need to be removed from an exhibit, pictures, descriptions, or three-dimensional printed copies can be substituted, and the augmented reality of visitor experience can adapt accordingly.

Presentation at the Department of History and Classics, Acadia University

  • Posted on: 5 March 2015
  • By: warren


Mapping the Western Front: the British and German experiences
March 26th, 7pm, BAC241
The static nature and scale of the battles on the Western Front was unwelcome to both Entente and Central powers during the Great War. Faced with logistical requirements on an unprecedented scale, standardized maps at different scales had to be produced of the battlefield quickly for both tactical and strategic purposes. This was a minor revolution in military thinking: previously cavalry officers were expected to ride with a sketch-board to map out terrain and enemy positions for their commanders.
In this talk I will contrast the Entente and Central efforts at mapping battlefields, highlight the differences in the approaches they took as well as evidence about local military intelligence activities. Both British and German coordinate systems will be explained as well as how to geo-reference these maps into modern mapping software.

Code Event: How to open a coffee shop in Halifax in 3 minutes

  • Posted on: 28 January 2015
  • By: warren


How to open a coffee shop in Halifax in 3 minutes
Code Event Halifax
Theater B, Tupper Building, Dalhousie University
January 30th, 2015, 7PM
In this talk I will demonstrate the power and value of open data by showing how an entrepreneur can choose a location to open a new coffee shop in Halifax using data-sets available on the Halifax and Canada Open Data Portals. Specifically, we will locate an available rental space for the coffee shop while keepings in mind the locations of potential competitors and customers.
Slides are available here.

Invited Talk: "Fear, uncertainty and doubt: doing good research work with partial network data"

  • Posted on: 29 September 2014
  • By: warren


Invited Talk: "Fear, uncertainty and doubt: doing good research work with partial network data"
NEH Workshop on Digital Methods for Military History
10-11th October 2014
As digital methods and tools become more prominent in the scholarly historical community and the practice of public history, historians have to learn to use these methods effectively. Northeastern University's NULab for Texts, Maps, and Networks, announces a workshop designed to help historians of the military to learn about and use these tools. This workshop is a partnership between the NULab for Texts, Maps, and Networks at Northeastern University, the NEH's Office of Digital Humanities, and the Society for Military History.
This two-day workshop will introduce historians of the military and foreign policy to digital tools and methods, focusing on network analysis and mapping, which are methods particularly suited to the study of the military. Participants will learn about projects that have successfully used these methods, and then they will receive hands-on instruction to help them get started in using these methods themselves.
Abby Mullen giving a talk on the uses of the Map Warper with historical maps.

rectifi.jpg panel.jpg




Panel on the second day on the uses of digital mapping techniques and uses.

Presentation at TSIO: 3D virtual worlds as search, discovery and retrieval engines

  • Posted on: 9 September 2014
  • By: warren


3D virtual worlds as search, discovery and retrieval engines
Thursday, 11th September 2014 - 4PM, City University London, College Building, Room A214
Joint Paper with David Evans, University of Derby.
The search for "relevant" information has long been driven by keyword searches and some basic visualizations.  The increase in both the amounts of data available and in the breath of data that is not a discrete text document is creating new opportunities for non-traditional means of information retrieval (IR).  In this paper, we present a prototype system where a 3D virtual world is used to access and discover information in semantic web databases. A prototype that focuses on the period of the Great War is discussed.