Richard Susskind on AI for Lawyers: A Review of 'How to Think About AI'

In his newest e book, Richard Susskind’s purpose isn’t to promote hype or pour chilly water on all the pieces — it’s to switch fuzzy impressions with crisp psychological fashions. Make sense of AI’s influence on the authorized occupation. Our assessment explains how Susskind’s e book supplies readability for perplexed attorneys.

Richard Susskind is what I name a Large Thinker — the type of authorized expertise sage who delivers keynote speeches by the dozen and publishes deep ideas in critical journals. He has a uncommon expertise for synthesizing complicated concepts and explaining them in a transparent and interesting means. For attorneys and policymakers making an attempt to make sense of AI’s influence, ”Find out how to Assume About AI: A Information for the Perplexed” could also be Susskind’s most well timed and clarifying work but.

“Find out how to Assume About AI: A Information for the Perplexed” by Richard Susskind. Oxford College Press, 2025. 224 pages. A brief, non-technical information that challenges us to assume in another way about AI.

Obtainable from Bookshop.org (helps unbiased booksellers), Barnes & Noble and Amazon.

Changing Fuzzy Pondering with Clear Fashions

This isn’t a e book of solutions. It’s a e book of questions. Susskind doesn’t fake to have options to all thorny points round synthetic intelligence. His purpose is to assist readers establish the problems and recommend some clear methods of desirous about them. Susskind’s purpose isn’t to promote hype or pour chilly water on all the pieces; it’s to switch fuzzy impressions with crisp psychological fashions. 

Readers bored with AI discourse that both cheerleads all the pieces or condemns it wholesale will seemingly discover the creator’s considerate and balanced strategy refreshing.

Susskind makes complicated concepts accessible to each lay readers and specialists. The e book does double obligation as a confidence booster for laypeople and a actuality verify for insiders who could also be over- or under-reacting to current breakthroughs. Cramming all this into 224 pages is sort of a feat. 

Key Ideas in Find out how to Assume About AI

Chapter 1, “The Summer time of AI,” is a brisk abstract of AI’s previous, current, and believable futures. Susskind sketches the early goals of laptop scientists, the winters when funding dried up, the deep-learning renaissance of the 2010s, and the present period of generative fashions. From there he invitations readers to think about the place issues would possibly go subsequent: self-improving techniques, autonomous researchers, or instruments that seamlessly mix into on a regular basis life. This historical past lesson isn’t simply trivia. It’s used to point out repeating patterns — overpromising, backlash, regular progress — that assist us spot what’s actually new this time.

Chapter 2, “On Expertise,” incorporates a warning:

We’re nonetheless warming up. In not a few years, our present applied sciences will look primitive, a lot as our Nineteen Eighties equipment seems antiquated at present. [The current wave of AI apps] are our faltering first toddler steps. Most predictions concerning the future are for my part irredeemably flawed as a result of they ignore the not but invented applied sciences.

He notes that Ray Kurzweil’s “legislation of accelerating returns” seems to be coming into play: “Data applied sciences like computing get exponentially cheaper as a result of every advance makes it simpler to design the subsequent stage of their very own evolution.”

These are in all probability the explanation why even high laptop scientists, together with Stephen Wolfram, can’t clarify how generative AI works. Susskind quotes Wolfram: “It’s difficult in there, and we don’t perceive it — despite the fact that ultimately it’s producing recognizable human language.”

The implication? We’re not simply on a brand new highway — we could also be constructing a brand new type of automobile whereas already driving at velocity.

The “course of vs. end result” distinction.

Chapter 3, “Course of-thinking and Final result-thinking,” units the stage for following chapters by contrasting the views of two heavyweight public intellectuals: Henry Kissinger and Noam Chomsky. Kissinger praises AI to the heavens. Chomsky thinks it’s principally nugatory. Susskind’s rationalization for the contrasts is that Kissinger focuses on outputs, whereas Chomsky focuses on course of:

Course of-thinkers are involved in how complicated techniques work.

Final result-thinkers have an interest within the outcomes they bring about.

Course of-thinkers have an interest within the structure of techniques.

Final result-thinkers focus on their operate.

Final result-thinkers additionally are typically “bottom-up” thinkers, preoccupied with general influence.

It is a key distinction that explains loads about variations of opinion about AI. Since AI apps don’t assume the best way people assume, process-thinkers are inclined to dismiss them as ineffective. Final result-thinkers are extra pragmatic, specializing in the demonstrable sensible advantages. They perceive that “machines don’t want to repeat us to ship the outcomes or outputs that clients, shoppers and customers need from their suppliers.” Attorneys, educated to research course of, could also be predisposed to dismiss AI’s unfamiliar logic, lacking the forest (helpful outcomes) for the timber (alien strategies).

Diving Additional Into Ideas

Subsequent up, Chapter 4, titled “Confusions,” explains some AI fallacies. This consists of how process-thinking can contribute to myopia about AI and its results, together with “Not-Us” considering:

Professionals see a lot higher scope for AI in disciplines aside from their very own. Medical doctors are fast to recommend that AI has nice potential in legislation, accounting and structure, however instinctively they have an inclination to withstand its deployment in well being care. Attorneys assert confidently that audit, journalism and administration consulting are ripe for displacement however supply particular pleadings on its very restricted suitability within the observe of legislation and the administration of justice.

Chapter 5, “We Don’t Have the Phrases,” makes the purpose that speaking about AI requires a brand new vocabulary. For instance, AI skeptics are fond of claiming that computer systems can by no means substitute them as a result of they don’t have the identical judgment, empathy and creativity. 

What’s missed is that computer systems can present what we would name quasi-judgment, quasi-empathy and quasi-creativity. Susskind demonstrates — fairly convincingly — that the pc variations of those organic traits may be superior in a number of methods to the human model. Skeptical about this? All I can say is try Chapter 5 earlier than changing into too assured that you’re irreplaceable.

In Chapters 6 and seven, Susskind considers how AI would possibly change the office. This consists of distinguishing three ideas:

Automation (process substitution) is about discovering a option to be extra environment friendly about doing what we’re doing now.

Innovation means delivering the outcomes shoppers need, utilizing methods or expertise that assist radically new underlying processes.

Elimination means not simply fixing an issue however eliminating it.

Many analysts see automation as the largest AI menace to jobs. Innovation and elimination could also be larger risks. Susskind makes a convincing case — to me, at any fee — that utilizing AI to implement totally different approaches to battle decision or prevention of authorized issues may scale back or substitute litigation as we all know it.

Sisskind’s seven classes of AI threat.

Chapters 8 and 9 take care of AI dangers and the way to tackle them. Quite than have interaction in unfocused handwringing concerning the dangers of AI, Susskind organizes his evaluation utilizing a easy desk:

CATEGORIES OF AI RISKCategory 1: Existential RisksThreats to the long-term survival or potential of humanity.Class 2: Dangers of CatastropheLarge-scale disasters or societal disruptions wanting extinction.Class 3: Political RisksImpacts on democracy, governance, surveillance, and autonomy.Class 4: Socio-Financial RisksEffects on employment, inequality, social cohesion, and bias.Class 5: Dangers of UnreliabilityIssues arising from AI errors, inaccuracies, or “hallucinations.”Class 6: Dangers of RelianceDangers of over-dependence or inappropriate belief in AI techniques.Class 7: Dangers of InactionNegative penalties of failing to develop or deploy helpful AI.

Having laid out the dangers, Susskind supplies options for coping with them. He emphasizes measured urgency somewhat than end-of-the-world hysteria. His message is that policymakers and the general public want to know the scale and velocity of present AI shifts, not as a result of catastrophe is inevitable, however as a result of choices made within the subsequent few years will ripple for many years. The subtext: Burying your head within the sand isn’t a impartial act — it quietly fingers the steering wheel to whoever is paying consideration.

The ultimate three chapters tackle philosophical concepts and hypothesis as to what the longer term might maintain for AI — and humanity. Discussions of Plato’s allegory of the cave, umwelten and Kant’s distinctions between phenomena and noumena more than likely gained’t have interaction the eye of each lawyer, however Susskind’s conclusion more than likely will:

My guess is that now we have not less than a decade to resolve what we wish for humanity after which to behave upon that call — if needed, emphatically and pre-emptively — by nation and worldwide legislation. [O]ur future will rely largely on how we react over the subsequent few years.

What Does All This Must Do with Attorneys?

ABA Formal Opinion 512 interprets summary considerations about AI into concrete moral obligations for attorneys, demanding competence in understanding AI’s advantages and dangers (Mannequin Rule 1.1), diligence in defending shopper confidentiality (Mannequin Rule 1.6), readability in shopper communications (Mannequin Rule 1.4), candor towards tribunals (Mannequin Rule 3.3), efficient supervision of AI use (Mannequin Guidelines 5.1, 5.3), and reasonableness in charges (Mannequin Rule 1.5).

Whereas not an ethics compliance handbook, Susskind’s e book affords exactly the conceptual instruments — the “psychological fashions” — wanted to navigate these sensible obligations. For instance:

Susskind’s discussions on AI capabilities, limitations, and the problem in explaining how some techniques work (Chapters 1, 2 and 5) immediately inform the obligation of competence below Rule 1.1, which requires attorneys to know the advantages and dangers of related expertise. 

His structured evaluation of AI dangers (Chapters 8 and 9) supplies a framework for assessing potential threats to confidentiality below Rule 1.6, significantly regarding knowledge safety and inadvertent disclosure when utilizing third-party AI instruments. 

Exploring the “course of vs. end result” distinction (Chapter 3) can illuminate challenges in speaking AI use to shoppers (Rule 1.4) or guaranteeing candor to tribunals (Rule 3.3) concerning the origins and reliability of AI-generated supplies. 

The worth proposition of Susskind’s e book lies considerably in equipping attorneys with the cognitive framework essential to operationalize the moral necessities newly formalized in Opinion 512. 

Don’t Get Left Behind: Learn Susskind on AI

With over 500 generative AI apps for attorneys cataloged by LegalTech Hub in March 2025, the proliferation of such instruments reveals no indicators of slowing. On this atmosphere, nuanced understanding is extra vital than merely one other utility. What the world wants is obvious explanations of the methods AI is altering our world now — and what we will anticipate tomorrow. 

Whether or not you’re writing briefs, litigating high-stakes issues, lobbying policymakers or simply making an attempt to future-proof a profession, Susskind’s e book goals to provide you adequate readability to steer somewhat than drift. And within the AI period, that is perhaps probably the most sensible present of all.

“Find out how to Assume About AI” is the literary equal of a well-lit commentary deck overlooking a stormy sea. It’s as a lot about society, ethics and identification as it’s about neural networks. For attorneys plotting technique in a generative-AI world, this e book is required studying.

“Find out how to Assume About AI: A Information for the Perplexed” by Richard Susskind. Oxford College Press, 2025. 224 pages. Obtainable from Bookshop.org (helps unbiased booksellers), Barnes & Noble, and Amazon.

Picture © iStockPhoto.com.