Digital Economy Dispatch #043

4th July 2021

Not Subscribed Yet? Sign up Here!

Why AI is Neither Artificial nor Intelligent: Discovering the Dragons in Kate Crawford’s “The Atlas of AI”

Any sufficiently advanced technology is indistinguishable from magic.
Arthur C. Clarke

There is something fascinating about looking at old maps. The bright colours and out-dated language are always appealing. And for the older maps, you get a sense from the mis-drawn boundaries of how much was unknown, how territories changed names and affiliations as political regimes changed, and how different areas evolved over time as the land was populated or developed.

And of course, the most fun of all to be found on some of the oldest maps is to see the parts of the world that were least known and simply labelled: “There be dragons!”.


To many people, AI must seem like magic. Presented with a new, unseen situation, a computer program is able to make sense of what’s happening and decide whether what is in front of you is a cat or a cow, whether to apply the brakes on your car or accelerate when the vehicle ahead of you swerves into your lane, whether the blood sample being analysed indicates the disease requires treatment or is benign, whether the person on the plane in seat 14C is acting suspiciously or just nervous about his first flight.

What could be better than a world where such technologies increase the speed, accuracy, and effectiveness of our decision making to drive business growth, improve our lives, and make society fairer and more equitable? Yet underneath this promise lies a more difficult reality. As we advance technology toward creating an “Artificial Intelligence (AI)” able to support these needs, the basis on which it operates needs to be open, honest, and transparent to avoid mistakes and expose manipulation.

In Kate Crawford’s new book, “The Atlas of AI”, she explores the AI landscape to offer a perspective on the history, context, and primary mechanisms that underlie this approach. Fuelled by large amounts of data, she openly questions if the individuals and organizations responsible for the development and promotion of AI are adequately considering where this data comes from, how it has been assessed, who controls and manages its use, and whether it is representative of the communities it is intended to serve. With a wealth of examples and illustrations, she brings to life the reasons why current AI practices pose important challenges for all of us interested in ensuring digital advances serve the broader needs of society.

Yet, more than a focus on the mechanisms and practices of AI, Crawford take a much deeper route through this landscape. She challenges the fundamental concepts of AI portrayed by many as a mythical bringer of insight and an infallible source of automation, efficiency, and productivity. Indeed, the very term “artificial intelligence” implies an other worldliness, a bringer of insights beyond our current human capabilities, and an expression of hope that we are able to extend our understanding of reality. Is that a fair reflection of how the AI industry operates today?

In her many interviews about the book, Crawford’s position is summarized in the phrase: “AI is neither artificial nor intelligent”. More substantively she raises 2 important points. First, that AI is not an abstract notion but directly derived from intense human activities to power its growth and intimately connected to the human condition through its influence, impact, and insights. AI is nothing but a human experience. All the imagery and jingoism of the tech world obscures the humanness of AI.

Second, that AI is not founded on a divine understanding of the nature of the world but built on human-defined rules, categorizations, and pattern matching techniques requiring huge amounts of data to be analyzed by vast computing engines. The intelligence, as such, is little more than a hyper-powered extension of all the human foibles and frailties that we must deal with in today’s society. As such, addressing the challenges of the murky history, biases, and political influences that plague other walks of life is just as essential in the field of AI as in every other area of our lives.

It is an enjoyable, if rather sobering book. It is at its best when it is focused on telling us more about what lies within the belly of the beast, exploring the hidden underpinnings of the AI industry and shining a light on the processes and practices that guide the application of the data-driven mechanisms at the core of the approach. Much like Andrew Blum’s book “Tubes” took us on a tour behind the scenes of the Internet, Crawford offers a detailed review of the sources of data that fuel the AI revolution and have become the subject of exploitation, bias, misdirection, and, let’s face it, fraud.

How would I sum up the book? In a single word: Relentless. The narrative of the book is simple and clear. But the detailed discussions and examples feel like an onslaught. Rather than clarifying and insightful, by the time I was half way through the book I was overwhelmed and looking for a way out. This is unfortunate because this is an important book on a critical topic of our time. By turning over the rock, Crawford is aiming our attention at questions and issues that many have wanted to avoid: The moral and ethical obligations of the development and use of digital technology.

But don’t let that put you off. This is a book that adds to the debate about the future of technology and our society. It will help you to ask better questions about what we’re willing to compromise to advance toward a digital economy and it will make a lasting difference to how you will view the nature and future of AI.

Digital Economy Tidbits

Get Ready for Years of New Tech Regulation. Link.

Maybe it is not just death and taxes we have to look forward to…..add to the list “lots of new tech regulations”. Sigh.

It’s not just antitrust actions that are coming, Microsoft president Brad Smith said in an interview. He sees new laws coming on everything from privacy and AI to cybersecurity and sustainability.


Data Saves Lives. Link.

Based on lessons from the past 18 months, here is the NHSx data strategy for how to make data more accessible for health care use.

The strategy sets out the Secretary of State’s vision for how data will be used to improve the health and care of the population in a safe, trusted and transparent way. It provides an overarching narrative and action plan to address the current cultural, behavioural and structural barriers in the system with the ultimate goal of having a health and care system that is underpinned by high quality, readily available data. It marks the next steps of the discussion about how we can best utilise data for the benefit of patients, service users, and the health and care system.

Not yet subscribed?

Sign up now to received Digital Economy Dispatches in your mailbox…..