Artificial Intelligence (AI) & VisiRule
People often ask 'What is AI?' and 'Where is the AI in VisiRule?' and 'Can I see the AI?' This page attempts to answer these questions.
What is AI?
You may be thinking that Artificial intelligence (AI) is some new area of computing, based on Deep Learning, which plans to build intelligent machines to take over from humans? Wrong.
AI is not new!
AI is not new -- it's been around for over 50 years. AI is not likely to replace humans in the foreseeable future and, moreover, AI is a 'broad church' comprising many complex technologies and techniques.
Machine Learning and Deep Learning are just two strands of AI research which happen to be very popular and fashionable right now. As to AI's potential usage, that is for philosophical and political debate.
AI is dumb, stupid!
AI is dumb; but even dumb things can appear smart and to do 'clever' things.
Is AI an Oxymoron?
Hard to say -- but AI is certainly more 'artificial' than 'intelligent' at present.
I know what I know - Right?
Not necessarily - you may know how to say fix something, but that does not mean you know how to express that knowledge in a coherent way. This is one of the big problems with building expert systems. We need the subject matter experts to be able tp specify their knowledge in an organized and unambiguous way, Having access to a soft, visual tool such as VisiRule is a great aid in achieving this, as ideas can be rapidly drawn and tested.
How can we define Artificial Intelligence?
AI is something which mimics the jobs normally associated with humans and appears to display intelligence
AI is something which is internally constructed in a way which supports or reflects human reasoning
VisiRule, along with other products in the LPA software suite, satisfies the above criteria. But, we don't really understand what 'intelligence' is.
AI as a Capability
Some people argue that AI is not so much a thing as a capability. AI enables software to look or appear to have:
Artificial Intelligence Timeline Infographic – From Eliza to Tay and beyond (courtesy to syzygy).
The AI Effect
Artificial Intelligence has always had an identity problem. AI often refers to ideas and techniques which haven't as yet been fully discovered. Once they emerge out of the bag of tricks known as AI, they often get their own distinct name such as Vision Recognition, Language Processing and are no longer attributed to AI. This is known as the AI Effect.
John McCarthy, the Father of AI, famously said: "As soon as it works, no one calls it AI any more."
Leading researcher Rodney Brooks says "Every time we figure out a piece of it, it stops being magical; we say, 'Oh, that's just a computation.'"
Nick Bostrom, Director of the Future of Humanity Institute at Oxford University states: "A lot of cutting edge AI has filtered into general applications, often without being called AI because once something becomes useful enough and common enough it's not labelled AI anymore."
Machine Learning offers the potential to learn how to predict future actions by analysing historical data. Intelligent algorithms can be used to train software to recognize, detect and classify certain situations. A lot has been written about ML especially Deep Learning but to make it work you need a LOT of good, clean, and UNBIASED data and a lot of processing power. Machine learning can help spot or diagnose problems which are about to occur, but can not help determine what to do about it, nor explain How or Why it reached this conclusion.
As Bill Ruh, CEO at GE Digital and chief digital officer at GE states: "AI will never be able to give the user the correct answer to the question of what to do next. The only way to do it is through modeling and simulation, looking at every instance of corrosion to understand the physics of what is happening to the pipeline. AI doesn’t understand this physics; it understands patterns.”
You can read what Forbes has to say about Machine Learning and Deep Learning and Expert Systems here.
AI Expert Systems use rules to replicate the behaviour of human experts. Rules can work forwards i.e. from data to conclusions; often called data-driven, or backwards i.e. from conclusions to data; often called goal-driven.
Humans are good at doing things --- they often have years of experience in doing something and have learnt to recognize and detect what to do in certain situations. People have the ability to make jumps in analysis and link in information outside of the box.
VisiRule allows us to combine machine learning and human knowledge in a simple and uniform way. VisiRule charts can be derived from data using Decision Tree learning and classification algorithms, and then reviewed and refined by human experts who can then incorporate their know-how.
It is interesting to note, harking back again to the exponential growth of information technology, that the hardware on which Watson ran in 2011 was said to be about the size of the average bedroom. Today, we are told, it runs on a machine that is the size of three pizza boxes, and by the early 2020s Watson will sit comfortably in a smartphone."
― Richard Susskind, The Future of the Professions:
How Technology Will Transform the Work of Human Experts
Expert Systems & Big Data
Within the landscape of AI, within the symbolic, rules-based cluster, are Expert Systems. In "Best Practices to Building an Expert System", John Etherington explains how 50-year-old AI innovation may solve big data problems.
Expert Systems look to provide advice and guidance of a quality and consistency comparable to that of a suitably skilled and experienced human expert. An expert system is an example of a knowledge-based system, where rules are used to represent the knowledge of the expert, rather than embedded in formulae or code. The goal of knowledge-based systems is to make the critical information required for the system to work explicit rather than implicit. This means that the expert knowledge or 'know-how' is more easy to identify, discuss, refine, revise and extend. It also means that systems built on this knowledge can use that same knowledge to explain how a conclusion was reached, as opposed to Neural Nets which can not explain how they arrived at any given conclusion.
In "Expert Systems, Artificial Intelligence and the Behavioural Co-ordinates of Skill", H. M. Collins classifies expert systems into four levels beginning with computerization of a rule book, followed by the incorporation of heuristics obtained by interviewing experts but used by humans only as an adviser, followed by expert systems acting autonomously and, finally, by systems with common sense. VisiRule is at level two with plans for level three. Obviously, there's a very big gap between levels 3 and 4.
Rule-based systems are used as a way to store and manipulate knowledge to interpret information in a useful way. Rule-based expert systems have long been associated with AI and the provision of a dedicated data structure to model human reasoning is one of the characteristics of AI.
The area of Knowledge-based Systems and rule-based inference falls under what is now sometimes referred to as "Good Old Fashioned Artificial Intelligence", or GOFAI, and is often characterized by symbolic reasoning, non-deterministic search and meta-level reasoning. VisiRule supports all of these, but chooses to present a 'simple' story using a familiar mechanism, namely the flowchart, which in VisiRule also resembles a decision tree.
The recently published Government by Algorithm report states: "Artificial intelligence (AI) promises to transform how government agencies do their work. Rapid developments in AI have the potential to reduce the cost of core governance functions, improve the quality of decisions, and unleash the power of administrative data, thereby making government performance more efficient and effective. Agencies that use AI to realize these gains will also confront important questions about the proper design of algorithms and user interfaces, the respective scope of human and machine decision-making, the boundaries between public actions and private contracting, their own capacity to learn over time using AI, and whether the use of AI is even permitted. These are important issues for public debate and academic inquiry."
Different Types of AI
Narrow vs General AI
The White House recently published: "Preparing for the Future of AI" report which stated:
"Remarkable progress has been made on what is known as Narrow AI, which addresses specific application areas such as playing strategic games, language translation, self-driving vehicles, and image recognition. Narrow AI underpins many commercial services such as trip planning, shopper recommendation systems, and ad targeting, and is finding important applications in medical diagnosis, education, and scientific research. These have all had significant societal benefits and have contributed to the economic vitality of the Nation."
"General AI (sometimes called Artificial General Intelligence, or AGI) refers to a notional future AI system that exhibits apparently intelligent behaviour at least as advanced as a person across the full range of cognitive tasks. A broad chasm seems to separate today’s Narrow AI from the much more difficult challenge of General AI. Attempts to reach General AI by expanding Narrow AI solutions have made little headway over many decades of research. The current consensus of the private-sector expert community, with which the NSTC Committee on Technology concurs, is that "General AI will not be achieved for at least decades."
Academic AI vs Pragmatic AI
Historically, there has been a disconnect between the theoretically inclined academic community and the more pragmatic industrial researchers looking at AI. The academics have been almost trying to solve the complete challenge of understanding, modelling and replicating human conscious thought processes, whereas the pragmatists have just tried to use AI techniques to solve specific challenges. A good example is Watson by IBM, a computer program which was designed solely to beat the world's leading players in Jeopardy and succeeded. Whilst Watson was never intended to do any deep semantic analysis or attempt to derive the meaning behind questions, it is now be heralded as the 'answer' to many questions in law, healthcare and others.
LPA are pragmatists. VisiRule strives to provide a transparent solution for delivering intelligent applications using both existing data and the knowledge of human experts, be they legal, medical, electrical or whatever.
Hard AI vs Soft AI
The ABA states in 'How artificial intelligence is transforming the legal profession': "There are two types of artificial intelligence—hard and soft. Hard AI is focused on having machines think like humans, while soft AI is focused on machines being able to do work that traditionally could only be completed by humans. The main difference is that soft AI doesn’t necessarily involve machines thinking like humans."
“Our perspective on artificial intelligence has changed significantly over the past several decades” says Jack Conrad, lead research scientist in corporate R&D at Thomson Reuters and president of the IAAIL. “AI failed to live up to the early expectations that focused largely on hard AI capabilities, such as the ability to perform human-like reasoning. When those lofty goals were not attained, researchers came to understand how difficult such achievements really were. After all, trying to teach computers to perform cognitive activities was an extremely challenging task. Over time, as expectations were lowered and research efforts became more narrowly directed, a shift towards ‘soft’ AI applications took place, focusing on providing intelligent tools and problem-solving resources to humans.”
A principal goal of VisiRule is to make it simple and easy-to-use, so that business users who understand their line of business can use it directly. Afterall, they hold the knowledge, and it is they that need help in extracting that precious knowledge and organizing it in a coherent and manageable way. VisiRule helps address this 'knowledge elicitation' problem, which historically has been the bottleneck in developing intelligent applications, by combining a visual model with rapid rule generation, instant compilation and immediate testing.
The key challenge is that of extracting and exposing knowledge which is buried within human brains, manuals, technical papers and transforming it into actionable rules to create automated systems. Experts do not know what they know, they often can not even explain how or why they came to a conclusion in a given situation. Concepts such as 'intuition' and 'instinct' are very, very hard to quantify and capture. VisiRule helps experts explore their own knowledge by providing a very soft and flexible framework.
VisiRule is a graphical tool which also plays the role of a rule generator. Rather than require the author define the rules using some rule language, the author simply draws a connected diagram which the VisiRule compiler translates into executable rules. The drawing task is helped by VisiRule knowing, to some extent, what the intended meaning and context is of each box as it is being drawn and/or linked. As the generated rule-base is executed, and questions presented, VisiRule can present a graphical view of the original chart and the active session.
The common wisdom about artificial intelligence is that we are building increasingly intelligent machines that will ultimately surpass human capabilities and possibly even threaten mankind. This narrative is both misguided and counterproductive. Framing AI as a natural expansion of longstanding efforts to automate tasks makes it easier to predict the likely benefits and pitfalls of this important technology.
Jerry Kaplan—CodeX Fellow and Visiting Lecturer, Computer Science, Stanford University
AI Material by Adrian Hopgood
"Artificial Intelligence for All" is an excellent introductory lecture by Prof Adrian Hopgood on YouTube which introduces the principles of AI with a focus on practical applications ranging from the control of manufacturing processes to the screening of mouth cancer.
LPA AI Technology Stack
LPA is a well established AI company with many years of experience in developing and deploying AI software solutions. LPA was recently recognized as a leading AI innovator by CV Magazine.
A Brief History of AI on AITopics detailing major events year by year
Artificial Intelligence – Application in Legal by Ben Gardner
AI Is Becoming A Major Disruptive Force In Banks' Finance Departments by Steve Culp (Forbes)
History of AI
Many people have written about the history of AI and its adoption over the years. Here are some 'good' ones.