Think Differently Because of AI

Ten books. Three thinking disciplines. One leadership agenda for responsible banking transformation.

The winning organizations will not necessarily be those with the most AI.

They will be those that know how to think differently because of AI.

Transformation experience has taught me that technology rarely fails only because of technology. It fails for the same three reasons, again and again: operating assumptions go unchallenged, institutional consequences are underestimated, or people are expected to adopt a solution they had no role in shaping.

AI will magnify all three mistakes.

That is why I have been studying ten influential books on artificial intelligence—not because I want to become an AI researcher, and not because leaders need to follow every new technology trend. I am reading them because AI is becoming a leadership question.

It will influence how companies compete, how customers make decisions, how risks emerge, and how boards govern increasingly intelligent institutions.

Understanding the technology is important. But understanding the technology alone is not enough. AI leadership is the ability to translate technological possibility into trusted enterprise outcomes.

From experimentation to institutional change

The first phase of enterprise AI focused primarily on experimentation. Can AI create content, summarize research, improve productivity, automate customer service? These experiments have helped institutions become familiar with generative AI and identify immediate opportunities.

But the next phase will be considerably more consequential. AI systems are beginning to reason across information, participate in workflows, use tools, recommend decisions, and operate with increasing autonomy. The conversation must therefore move beyond productivity

Leaders must now ask:

– Which consequential decisions should be delegated to AI, and where must human judgment remain decisive?

– How will AI change the economics of banking, and what new risks will intelligent agents create?

– How should boards govern technology that evolves faster than annual oversight cycles?

Ten books helped me examine these questions:

1. The Singularity Is Nearer — Ray Kurzweil

2. Co-Intelligence — Ethan Mollick

3. Competing in the Age of AI — Marco Iansiti and Karim Lakhani

4. The Coming Wave — Mustafa Suleyman

5. Life 3.0 — Max Tegmark

6. The Age of AI — Henry Kissinger, Eric Schmidt, and Daniel Huttenlocher

7. Genesis — Henry Kissinger, Eric Schmidt, and Craig Mundie

8. Prediction Machines — Ajay Agrawal, Joshua Gans, and Avi Goldfarb

9. AI 2041 — Kai-Fu Lee and Chen Qiufan

10. The Singularity Is Near — Ray Kurzweil (2005 original, above is the 2024 sequel)

These books do not present one unified view. Some are optimistic, others cautious. Some focus on today’s workplace; others examine superintelligence, geopolitics, and the future of civilization. Their disagreement is part of their value—and reading across these perspectives sharpened three questions I believe every leadership team must ask.

First-principles thinking: What assumptions should we challenge?

“The Singularity Is Near” and “The Singularity Is Nearer” challenge our instinct to think linearly. Ray Kurzweil argues that advances in computing, AI, biotechnology, robotics, and nanotechnology are compounding. Whether one agrees with his precise predictions or not, the leadership lesson is difficult to ignore: technology may change faster than institutions can adapt. The objective is not to predict the future perfectly. It is to build institutions that remain adaptable when today’s assumptions become obsolete.

“Prediction Machines” offers a complementary economic lens: AI lowers the cost of prediction. That suggests a better executive question. Instead of asking “Where can we use AI?” ask “Where is uncertainty preventing us from making a better decision?” In banking, that applies to credit, fraud, liquidity, and operational risk.

But prediction creates no value by itself. Value comes from the complete decision loop: Prediction → Judgment → Action → Outcome → Learning. The unit of transformation is not the AI model. It is the consequential decision.

“Co-Intelligence” brings first-principles thinking into everyday knowledge work. Ethan Mollick describes AI as a new form of collaborator: capable, creative, inconsistent, and rapidly improving. It may perform brilliantly on a complex assignment and then fail on something apparently simple—so leaders cannot understand this frontier through presentations or delegated experimentation alone. We need direct experience, to discover where AI improves human capability, where it introduces hidden risks, and where human expertise remains indispensable. The principle I take from the book is straightforward: invite AI into the work, but never outsource accountability.

Many organizations are applying AI to workflows designed for a pre-AI world. The initial question is often “How can AI make this process faster?” First-principles thinking asks: “Why does this process exist in its current form—and would we design it this way if intelligent digital capability had always existed?” In banking, this means questioning why customers navigate disconnected products and channels, why decisions move sequentially between functions, why information is repeatedly collected rather than intelligently reused, and why standardized products persist when customer needs are increasingly knowable.

The objective should not be to automate yesterday’s complexity. It should be to identify the fundamental customer, business, and regulatory need—and redesign from there.

Systems thinking: What consequences and feedback loops will change?

“Competing in the Age of AI” explains why isolated AI use cases are insufficient. Sustainable advantage comes from an operating model that continuously converts data into intelligence, decisions, actions, and learning. A bank will not become AI-powered by placing separate AI solutions inside every existing product and functional silo; the larger opportunity is reusable intelligence across products, operations, risk, service, and customer journeys. That requires trusted data, modular architecture, redesigned processes, embedded governance, and continuous feedback.

A successful AI use case is not automatically a successful enterprise outcome. In banking, it becomes enterprise transformation only when it changes the quality, speed, accountability, or economics of a consequential decision.

“The Coming Wave” adds the questions of power and control. AI is becoming more capable, autonomous, affordable, and accessible simultaneously—creating extraordinary opportunity, but also enabling cyberattacks, automated fraud, and increasingly autonomous systems. For financial institutions, AI risk cannot be treated as an extension of conventional technology risk. Safety must become part of product design, deployment, and executive accountability.

“Life 3.0” extends systems thinking into alignment and control. An intelligent system does not need to be hostile to create harm—it may simply optimize the wrong objective extremely effectively. If we optimize approval speed, what happens to risk quality? If we optimize productivity, what happens to employee judgment?

Systems thinking prevents leaders from confusing local improvement with enterprise success. It asks: what other parts of the institution will this affect, which feedback loops will it create, where will work, risk, power, or accountability move, and what happens when the solution operates at scale? AI leadership requires us to optimize the institution—not merely the individual use case.

Responsible innovation as a source of advantage

In industries built on confidence and trust, responsible AI is not an obstacle to innovation. It is a condition for sustainable innovation—and increasingly a source of advantage. Responsible AI capability improves speed, because a bank that can explain and evidence its models moves through governance and regulatory review faster. It improves access, because trust-sensitive segments—wealth, corporate, government relationships—reward institutions that can demonstrate control over consequential decisions. It strengthens talent, because serious governance is where the best builders choose to work. And it improves resilience, because institutions that have already built explainability, monitoring, and human override into their systems absorb the next model failure or regulatory shift without a crisis.

These benefits rarely show up in a single use-case business case. They show up in enterprise value over time. Leaders who treat responsible AI only as a cost are underpricing what it can become: a trust moat that is harder for competitors to copy than any single model or feature.

Design thinking: What human outcome should we create?

“AI 2041” makes potential futures tangible. Its scenarios remind us that technology never creates only first-order effects: AI healthcare changes access and professional roles, autonomous vehicles change employment and liability, synthetic media changes creativity but also identity and trust. For leaders, scenario planning must go beyond asking whether a technology will work—we must ask what happens after it works.

“The Age of AI” and “Genesis” move further into questions of knowledge, authority, and human purpose. A system may produce an accurate result without offering an explanation a customer, employee, regulator, or director can meaningfully understand. Is accuracy alone sufficient? Who is accountable when an algorithm influences a consequential decision? How much authority should an institution delegate to a machine, and how do we prevent human judgment from becoming ceremonial rather than real? My conclusion is that trust must be designed alongside intelligence.

Human agency should not remain an abstract principle. In practice, it means:

– People know when they are interacting with AI.

– Consequential decisions can be explained appropriately, and affected people can question or appeal them.

– An accountable human can intervene when necessary.

– The institution remains responsible for the outcome.

Design thinking begins with a genuine human need rather than the availability of new technology. The most important question is not “What can the AI do?” It is “What should the customer, employee, or decision-maker be able to achieve?” Technically successful AI may still fail if people cannot understand it, trust it, challenge it, or use it effectively. Design thinking turns AI from a technology deployment into an experience people can understand and trust.

The three frames work together

These disciplines should not be applied independently. First-principles thinking identifies what should be reconsidered. Systems thinking reveals the wider consequences of changing it. Design thinking ensures the solution works for the people it is intended to serve.

The leadership sequence becomes:

Challenge the assumptions.

Understand the system.

Design for the human outcome.

Experiment, learn, and scale responsibly.

This is how AI strategy becomes responsible transformation rather than technology-led
automation.

Institution, people, and decisions

Three outcomes anchor where this approach must ultimately create value:

– Institution: AI must strengthen the operating model, resilience, adaptability, and long-term competitiveness of the enterprise.

– People: AI should expand the capabilities of customers and employees without quietly removing their agency or right to challenge a consequential outcome.

– Decisions: AI creates enterprise value when it improves the quality, speed, and economics of decisions—while preserving clear accountability for their consequences.

These are interconnected. An institution cannot become more intelligent by weakening its people. Employees cannot exercise meaningful judgment if decision processes are opaque. And faster decisions do not create sustainable value if they weaken trust in the institution making them.

A board agenda for the AI era

AI oversight cannot be reduced to a periodic update on projects and model performance. Boards should be asking:

– Which consequential decisions are we prepared to delegate to AI, and where does management retain accountability?

– Are we measuring enterprise value or simply counting use cases?

– Can customers and employees understand, question, and appeal material AI-assisted decisions?

– How are we preparing the workforce to exercise judgment alongside intelligent systems?

These are strategy and legitimacy questions, not only technology questions.

My leadership agenda

These books and thinking disciplines reinforce four responsibilities for me.

1. Develop strategic foresight. Look beyond immediate use cases and understand how AI may reshape customers, competitors, ecosystems, and industry boundaries.

2. Translate possibility into an operating model. The objective is not to accumulate AI demonstrations. It is to build an institution capable of turning intelligence into trusted decisions and measurable outcomes.

3. Make responsible innovation a source of advantage. Governance, resilience, explainability, and human accountability should enable confident adoption—not arrive as a final checkpoint.

4. Preserve human purpose. AI should expand the capabilities of employees, customers, and institutions—not quietly remove their agency.

No single author can predict the future of AI with certainty. The optimistic forecasts may be early. The warnings may be incomplete. But waiting for certainty is not a strategy. The strongest institutions will combine intelligence with judgment, automation with accountability, innovation with trust.

For me, the leadership agenda is clear:

Challenge inherited assumptions through first-principles thinking.

Understand system-wide consequences through systems thinking.

Create trusted human outcomes through design thinking.

Then experiment, learn, and scale responsibly.

Where do AI initiatives most often fail in your organization: inherited assumptions, system-wide consequences, or human adoption?