Author: Silvie Spreeuwenberg

The book on eXplainable AI

Economists say that we need to keep the money moving for the economy to flourish. I strongly believe the same is true for knowledge. We need to continue to share knowledge for societies to flourish. My motivation to write a book about XAI, named AIX: artificial intelligence needs explanation, is to share my knowledge and encourage knowledge sharing within and between organizations.

The recent popularity of systems that claim to be artificially intelligent by having ‘machine learned’ knowledge gives the impression that even managing knowledge is a process that can be outsourced and automated. Every day I encounter opinions that say outsourcing knowledge creation to machines is scary. How do we control these machines? How dependent do we want to become and what does it mean for individuals, businesses and societies? Why is Henry Kissinger writing about a revolution driven by artificial intelligence? Why do the US, Europe and China compete in AI? What are the reasons that even the tech industry warns us about intelligent machines? All my life I have been working on improving knowledge management processes. I thought it was time to share my ideas on these topics with a wider audience.

You may be someone with a general interest in Artificial Intelligence – AI – or be a business manager. You see the potential of AI systems for your organization and want to stay in control. Mainstream businesses are researching the potential of AI and want to do that in a controlled way. This book will provide you with a strategy to achieve trustworthy AI systems. Prior knowledge about AI or system development is not needed to benefit from this book.

I studied AI and graduated with a thesis on generating a human readable explanation for a machine learned model. To my surprise, generating explanations is still not a standard feature of the currently popular AI platforms. Since most businesses had no big data when I graduated in 1996, I focused on decision support systems based on rules.

As a consultant, I worked mostly for large corporations. Typically, I operate in complex, knowledge intensive, environments: harbours, airports, insurance companies and tax administrators. I developed software, courses and methods to really engage business experts in making IT solutions. Many of the examples and inspiration to write this book are based on these experiences and the many interactions with business-people and other professionals.

I am grateful to many people who contributed to this book by discussing ideas, sharing experiences or reviewing the manuscript.

Machine learning technology trending

Why XAI?

Post author By Silvie Spreeuwenberg
Post date 18 July 2019

The five reasons why XAI solutions are more successful than an oracle based on AI, or any black box IT system, are as follows:

Decision support systems that explain themselves have a higher return on investment because explanations close the feedback loop between strategy and operations resulting in timely adaption to changes, longer system lifetime and better integration with business values.
Offering explanations enhances stakeholder trust because the decisions are credible for the customer and also makes the business accountable towards regulators
Decisions with explanations become better decisions because the explanations show (unwanted) biases and help to include missing, common sense, knowledge.
It is feasible to implement AI solutions that generate explanations without a huge drop in performance with the six-step method that I developed, and technology expected from increased research activity.
To be prepared for the increased demand for transparency based on concerns about the ethics of AI and the effect on the fundamental principles of a democratic society.

Each chapter in the book: AIX: Artificial Intelligence needs explanation, will explain one of these reasons in detail and provide examples or practical guidance. After reading you will have a good understanding what it takes to explain solutions that support or automate a decision task and the value explanations add to your organization.

I have tried to stay away from technical details. When I describe how to implement XAI solutions, I make an exception to this self-imposed rule and introduce some details about the underlying technologies. Just enough for a good understanding, not enough to ‘do it yourself’. In the final chapter I recommend sources to use if you need help understanding the algorithms in more detail.

design technology trending

User Interface Counts

Post author By Silvie Spreeuwenberg
Post date 7 June 2019

Since the invention of the computer, AI and IT have been closely related. Think of Lady Lovelace, who worked in 1837, together with Charles Babbage, on the first design of a programmable computer. Back then it was known as an ‘algebraic machine’ and she named it The Analytical Engine.

Lady Lovelace was the daughter of the romantic poet Lord Byron, and was a gifted mathematician and intellectual. When she translated an Italian article on algebraic machines, she supplemented it with extensive notes on such machine’s capabilities. These notes, published in 1843, prove that she was the first to record that a machine could be programmed to solve problems of any complexity or even compose music. This formed the inspiration for theories on logic that resulted in the programming languages in use today.

Lovelace died early and the Analytical Engine was not built during her lifetime. Her soulmate Mr. Babbage did create a trial model of the Analytical Engine that is displayed in the Science Museum in London.

She is regarded as the first computer programmer but was most likely thinking about AI, or what we would call AI today, when she said: “The Analytical Engine has no pretensions whatsoever to originate anything. It can do whatever we know how to order it to perform. It can follow an analysis, but it has no power of anticipating any analytical relations or truth.”

Her statement has been debated by Alan Turing, another example of a person in which the close relationship between AI and IT comes together. He defined a test, named after himself, of a machine’s ability to exhibit intelligent behavior. The test is an experimental setup. Someone judges a conversation without knowing whether the conversation is with a human or a machine. All participants are separated from one another. The Turing test is successful if none of the participants is able to tell the difference between communicating with the machine and communicating with the (other) human. This test, in many variations, still plays a role in defining AI.

At the same time Turing played a crucial role by creating one of the first computers based on the Von Neuman design: a computer that had a stored-program. This marked the beginning of programmable machines, the start of executing the vision of Lady Lovelace and the rise of the software industry.

The consequences of this innovation for humanity have been huge and were, at the time, difficult to oversee. There were pioneers, visionaries, investments and failures needed to get us to where we are today. I am so grateful with the result. Every day I use a smart phone to provide me travel advice, ways to socialize and recommendations on what to buy. Computers also help me memorize and acquire new knowledge. Many of these innovations are related to technology developed by researchers in Artificial Intelligence. The full potential has not yet been exploited.

Continue reading in the book: AIX artificial intelligence needs explanation.