Reimagining Government: Handy AI guide for govt, but falls short on detail

Reimagining Government:  Achieving the Promise of AI

by Faisal Hoque Erik Nelson and Thomas H Davenport and others

Published by Post Hill Press

249 pages  ₹2,521

  It is not often that one would recommend a book that is decidedly stodgy to read and contains no sweeping, grand insights. Yet this is an honest book, a prosaic book that seeks to do exactly one thing and succeeds quite well in the task.

 

The contents of Reimagining Government  are exactly what the title says. It is about how the government should use artificial intelligence (AI) to improve its functioning, provide better services to citizens, and to become more efficient. The volume has been put together by three co-authors — or perhaps authors and editors — whose names appear on the front cover. But the book credits five other contributors, all of whom have provided their expertise in putting together the book.

 

The volume they have produced is a sort of handbook or practical guidebook. The authors understand that many policymakers — both elected representatives as well as career bureaucrats — would not be very familiar with AI. Indeed, they assume that most of the policymakers at whom this book is aimed at would need a crash course in the simplest terms. This is what the first chapter — “Understanding the AI Tech Stack” — seeks to do, in terms that anyone, including AI phobes, would understand.

 

It is really chapters two to eight that cover everything from the uses of AI in governance to implementation plans and frameworks, the options available, why the government needs to partner with private players for AI implementation — and more importantly, the kind of organisation it should partner with. The chapters also discuss AI adoption and maturity curves, as well as the kind of government leadership required in the AI age.

 

The frameworks, the pros and cons of open versus closed AI models and systems, risk assessments and implementation challenges are some of the most useful areas covered.

 

The authors understand that too much information can be counter-productive to their audience and stick to the broad issues, without being superficial. The approach seems to be that policymakers need to understand both the big picture as well as have an understanding of the critical details but need not be an expert in all matters AI. Thus, most chapters present readers with the main issues that need to be debated and decided upon when choosing the partner for implementation, selecting the model that will work best for their particular use cases, and identifying the main issues to be scrutinised while designing the AI implementation framework and processes. Policymakers will obviously hire an expert — but he or she also needs to know enough to ensure that the expert does what is required. This means, he or she also needs to first understand what AI can and cannot do — and what is required for his or her department.

 

While the book is written obviously for policymakers in the United States, it can actually pretty well be used by most governments in both developed nations and fast-developing nations like India.

 

There are a few case studies that the authors use for explaining how AI has already achieved much in specific areas. Unfortunately, they are too concise. Also, more case studies would have been good. The authors could have looked at more countries for use cases of AI in government.

 

Chapter nine, which is on organisational redesign is a mixed bag. Like the previous chapters, it covers the basics in short. But because the areas covered are very complex — the evolving nature of AI-human relationship in government, for example — it would have merited a deeper, more nuanced examination.

 

The final chapter is both a summation as well as a continuation of the previous chapters , focusing on how to achieve the promise of AI.

 

One area that should ideally have been covered in some detail are the guardrails that are needed to be put by the government, both for preventing AI from going off the rails and creating havoc and to address societal risks. The authors do try to explore the main issues in chapter four, with the CARE framework, but it is dealt with in too little detail for something so important.

 

In India, NITI Aayog and others have published several white papers and strategy reports for the government on how best to implement AI. They go into great detail and may sometimes be intimidating for some people. This book deals with almost all the issues covered in several reports in an easy-to-digest form.

---

The reviewer is former editor of Businessworld and Business Today magazines, and author of Will India Get Rich Before It Turns 100? A Reality Check