How To Think About AI (Richard Susskind)
Published: 2026-01-04
A year is a long time in AI, and so when you pick up a book about AI there is already a fear that some of it might be out-of-date. Lucklily for us, Richard Susskind, Law Professor and AI thinker, has put sufficient thought into the topic of this book that I feel it will not be out of date any time soon. The shortest prediction timeframes begin at 2030, and so there are still a few years before we can even say that someone's prediction came to pass or not, and at the rate AI is evolving, that is a long time.
Firstly, you may ask: "How does a Professor of Law get to talk about AI"? If we take a look at Susskind's impressive credentials, not only did he write his theses on AI and Law, he is also the president of the Society of Computers and Law, and a Fellow of the British Computer Society. Susskind pioneered the use of AI systems in Law throught he 80s and 90s. So I'm happy to listen to what he has to say.
Key Ideas
The three key ideas that I took away from this book are:
- The 'AI Fallacy'
- Process-thinking vs outcome-thinking
- Three paths to value: Automation, Innovation, Elimination
It is difficult to rank these in order of importance, as they are arguably all perspectives on the same idea, that we tend to think about the impact of a new technology in our lives as an extrapolation of its current abilities and the abilities of humans.
The 'AI Fallacy' as Susskind describes it is the flawed thinking that:
the only way to get machines to perform at the level of the best humans is somehow to replicate the way that humans work
This thinking could be applied to various things through history. Whereas human inginuity has often come from nature and the behaviour of living organisms, it is not the case that this is always how we should think. In the case of AI, it is irrational to think that the best way to improve an existing process is to replicate that process digitally and improve on it. Quite rightly, many processes cannot be easily replicated, but their outcomes can still be achieved.
This is the second related point - that much of the current thinking is based around the process of achieving some goal, as opposed to the outcome of the goal itself. Consider automating the production of printed material - the process-thinkers of 100 years ago would have argued that the way to increased production was simply to create faster printing presses, but could not have imagined that such a machine could produce 350 sheets per minute (that's nearly 6 sheets per second). Yet that is what this printing press is capable off, and it definitely doesn't look like a faster version of Gutenberg's printing machine.
This is where Innovation takes over from Automation, which leads into the third key idea. Susskind argues that too many people focus on Automation, mainly because it is simply easier to think that way. This limits us from thinking about innovation - ways of achieving the same outcome with different tools or technologies. Taking this one step further, Susskind reminds us that with sufficient innovative thinking we can eliminate problems that required the original process in the first place. He uses the example of domestic chores (ironing in particular)... Thinking only in terms of Automation would mean building machines which performed the same ironing-actions with existing tools, which would be difficult to do. Innovation endowed us with the 'steam press' which achieves the same outcome in a different way, and further innovation resulted in 'wrinkle-free' materials which eliminates the need for ironing alltogether.
The Need for Thinkers
It is this way of approaching the opportunities and problems of AI which makes this book so important, and in fact important for thinking about any significant technology. For this reason, Susskind argues that there should be a co-ordinated international effort to articulate and discuss the future of this new technology, for one reason because those selected people should have the ability to think in this way.
Additionally, those currently in charge of the development of the technologies are either ill-equipped to think in this way or have a conflict of interest with the business value they are trying to accumulate. As he states in his intro:
They are often too vested in how and why they themselves to AI to speak reliably or objectively about its future and effects.
What should these thinkers discuss? Susskind spends sufficient time in the book covering the various hypothetical outcomes of AI, their multitude of risks and how to balance these by 'harnessing' AI. He does well in covering all the bases but without too much academic depth.
The Endgame
It is only toward the end of the book that things started to drift from the core topic, in my opinion. Two of the hypothetical outcomes of the adoption of AI are the 'singularity' (where human consciousness melds with that of AI), and 'AI Evolution' where humans are replaced by super-intelligent systems in a variety of different ways. These seem to be the realm of science-fiction, but are backed by some eminent and respected thinkers.
In reading the chapters, I felt that there was a need to include them as discussion points but Susskind treats them as integral to the book's climax and I think this takes away from the more important shorter-term thinking that the book calls for. Susskind decides to contemplate the purpose of human life in the cosmos and draws conclusions based on humanity's self-importance in the universe which I suppose can't be helped, but for me it strayed too far from the central topic and left me more perplexed (in contradiction tot he book's subtitle which claims to leave you less perplexed).
All in all, this is an important book for anyone, especially technologists, to read. It contains important ways of thinking that would benefit anyone across all professions which may or may not be impacted by AI. And think about it we must, as the next few years may prove crucial.