Non-Fiction Reviews
How To Think About AI
A Guide For The Perplexed(2025) Richard Susskind, Oxford University Press,
£10.99 / US$13.99, hrdbk, xiv + 192pp, ISBN 978-0-198-94192-7
While it's true that AI (Artificial Intelligence) seems to be front-of-mind for the general public and big companies alike right now, what is also true is that we have no idea of the impact that it is having, and will continue have, on all of us.
The tech companies at the heart of things tell us it will save time, cut costs, take over tedious, repetitive tasks, and give us more thinking/leisure time. The naysayers tell us it will take our jobs, out perform us and ultimately kill us all. The truth clearly lies somewhere between these two extremes, but what that truth is, no-one really knows.
Part of the problem is that, broadly speaking, it is difficult to explain what AI actually is and is not, what it can and cannot do, and whether or not it is good, bad or indifferent for us in the long term. The one solid truth in all of this though, is that we need to think about AI. Deeply. Beyond the legal, regulatory and compliance issues, there are big moral and ethical questions to be pondered, boundaries to be drawn and big decisions made about the AI future for humanity.
Richard Susskind has spent forty years thinking about AI. A lawyer by profession, he is now the world’s most cited author on the future of legal services and a leading expert on the impact of AI on society and this, his eleventh book, is his most general and definitely his most accessible: it is non-technical, written in a light and easy tone making it an easy read. What Susskind has done well is distil down the myriad of tomes written on the subject to an easily digestible paperback that discusses and then encourages debate on the pros and cons of AI. Personally, I would have liked a bit more detail here and there but in general its brevity is a strength that encourages you to keep reading.
He opens with a brief introduction and a history of AI, how long it’s been around, how we’ve been using it for years in ways we didn’t even know and the state of the art today with ChatGPT, before getting to the meat of the book.
Susskind posits that within AI there are two different kinds of thinkers. Process-thinkers, 'bottom-up' people pre-occupied with how complex systems work, their architecture and operational details, and outcome-thinkers, 'top-down' people interested only in the function, the results that complex systems bring and their overall impact.
It's not hard to work out that the tech companies are process-thinkers. They coo over algorithms, data sets, neural networks, machine learning and comprise Data Scientists and Software Developers. They want to know how the human brain works just so they can better mimic it with machines. Outcome-thinkers on the other hand wring their hands at the social consequences, economic impact, regulatory challenges and ethical implications; they are politicians, businessmen, lawyers and economists.
He talks about the ‘Not us’ professions, those that, amusingly, believe that years of study and professional experience somehow makes them immune to AI. Lawyers in particular fall into this category because they provide a service based on building up knowledge and experience. But clients have zero interest in a lawyer’s knowledge or experience, they just want a specific outcome – solve my legal problem. As Susskind points out, they are “...heavily invested in their years of education, training, and hard graft” with “...their status and often their wealth inextricably linked to their work.” Their focus is on processes not outcomes because they cannot envisage any other way of getting to the result.
Susskind believes AI will change our lives in three ways, automation, innovation and elimination. Automation is self explanatory, innovation is completely redesigning things we do to do them differently, but the most interesting is elimination, where AI advances might well produce something that removes a problem completely. The example he gives is how the problem of dealing with mountains of horse manure in the 1890’s wasn’t solved by dealing with the manure, it was solved by inventing cars. Our lawyers are just not capable of thinking like this, that “...the competition that kills you won’t look like you.”
Overall, Susskind stresses we need balance in our thinking but he leaves us with plenty to think about, acknowledging that we may already be too late, that even the experts don't really know how AI works now and expressing real concerns about the future. He claims we only have around ten years to decide if and how we can keep AI under control. He is adamant that we should be planning for AGI (Artificial General Intelligence) when machines will reach the level of human intelligence, and that it will happen within the next decade.
Regardless of which camp you may fall into, neither of these factions is inherently wrong, but for every incredible breakthrough in science, healthcare or engineering, there is a cost, be it social, economic or environmental, and that balance, weighing up the technological advances against the impact on humanity is how - and why - we should be thinking about AI.
Robert Grant
[Up: Non-Fiction Index | Top: Concatenation]