When most people talk about AI, they are talking about Generative AI (GenAI). GenAI is useful and valuable and will drive significant value, but the field of AI is much more than GenAI. AI has been under active development for over 70 years. Read all about the nuances and differences in each of the three areas of AI. 

As AI moves towards larger and more comprehensive applications and solutions, we will see an evolution of all three areas of AI, and, at the same time, we will see a convergence of the three areas toward Composite AI.

Composite AI will be the state of the art for many years and possibly even decades. Composite AI will grow and evolve into Artificial General Intelligence (AGI). While AGI is an exciting theme for science fiction movies, it will not arrive in the next few years. The path to AGI will be long and challenging. We discuss the pragmatic and practical path in detail.

Business leaders and technologists need to understand where AI is moving. We will outline one of the most probable paths unfolding over the next few decades as we move toward AI being embedded in all systems and operations. AI will become a utility like electricity and water, but we have a long road of sophisticated development to navigate before we arrive at that point.

The Path to AGI is a reference book and guide for those interested in all types of AI and how these types will merge, integrate, and evolve into one of the most consequential technologies the world has ever seen.

Unravelling the past, present, and future of artificial intelligence with pragmatism and precision, John Thompson’s “The Path to AGI” is an indispensable guide for business leaders seeking to understand the real capabilities, applications, and future trajectory of AI.

John’s decades of experience in analytics shine through as he navigates the Four Eras of AI, providing a roadmap to understanding the backdrop for today’s AI revolution and the long journey ahead to true Artificial General Intelligence.

Whether you are an industry leader, a technology enthusiast, or simply curious about how AI will shape our future, The Path to AGI is essential reading. Thoughtful, rigorous, and engaging, this book not only serves as a great reference but a lodestar for those navigating the ever-evolving landscape of AI.

Thomas Robinson, Chief Operating Officer, Domino Data Lab

This book has something for both the person already familiar with the development of AI and the novice just starting to learn about the field. In clear, narrative language, this book shows how AI has developed over time and the direction it is going. This would be a book I’d be put in front of students looking to be AI leaders in the future.

Cliff Lampe, Associate Dean for Academic Affairs at the University of Michigan School of Information

It is a bold act to write a book that describes the path to artificial general intelligence, but then John Thompson is a bold thinker. He doesn’t shrink from telling readers what’s going to happen with analytics, AI, data, and their implications for business and the world. I have generally found that he has been either correct in his predictions or close enough to be interesting and informative.

I must say, however, that this book contains his boldest predictions of all. When will machines become smarter than humans at every intellectual task, and how will it happen? What will be the implications of this achievement for humans and the planet? It’s difficult to imagine a more important topic.

This morning as I write in late February 2025, I listened to a podcast reviewing the achievements—or non-achievements—of the Paris AI Action Summit held earlier this month. Two topics discussed there are relevant to key AGI questions: when it will happen and what is the world going to do about it?

The heads of three of the world’s leading AI companies were in Paris, and on the question of when AGI would happen, they were all much more optimistic—if that’s the right word—than John Thompson. They predicted that AGI would be upon us in two or three years, or perhaps five at the most. I think this is unlikely, though my own prediction would be closer to theirs than to Thompson’s (I will let you read this book to find out what his prediction is).

That set of predictions from leading AI vendors is scary enough, but even scarier is what the assembled representatives of the world’s leading economies in Paris agreed to do about it—basically nothing. The new vice president of the United States, J.D. Vance, argued from the podium that regulation and AI safety were non-issues as far as the U.S. government was concerned (at least until 2029 when a new administration might take office; let’s hope that AGI hasn’t happened by then). Even some national leaders from Europe, where AI regulation is perhaps the furthest along in the world, expressed some concerns about excessive regulation stifling AI progress. At the end of the summit there was a mildly-worded declaration for attending countries to sign advocating inclusiveness, “multistakeholder” collaboration, and human-centric AI, but the U.S. and the U.K. even found those weak proposals objectionable and refused to sign the declaration.

This is disturbing because of the statement of John Thompson’s in this book with which I agree most wholeheartedly: “we are not ready for this.” We’re not ready for what AGI will do to jobs, education, creative expression, international relationships, human relationships, etc., etc. We don’t have regulation that even successfully addresses the current state of AI, much less AGI. We haven’t yet created general principles for how humans will work and learn and thrive alongside today’s AI. We have no global agreement on how even today’s AI should affect weapons and warfare. We don’t even have many leaders who understand today’s AI. We are woefully unprepared. I hope that John is correct that AGI is a long way off, because we are a long way off from being prepared to use it well and wisely. But I fear that he is not, and we don’t want to be complacent on the AGI timeframe topic if that leads to inaction.

I do agree with many of his fundamental principles for why AGI should take a long time to develop. The book is convincing and correct in my view that true AGI will need to be a composite of foundational AI (or what I would call analytical AI), generative AI, causal AI, and perhaps even symbolic AI. It’s worth reading this book just to get a perspective on all of these AI elements that need to come together. I also agree that while generative AI models may already appear to be generally intelligent, they aren’t won’t be until they have some causal and logical understanding of the world, as we humans do. AGI can’t result only from a set of statistical correlations and coefficients—no matter how much data they are based upon.

Even if you believe that it will require many decades to perfect and assemble all of these AI capabilities to achieve AGI, it’s important to begin thinking hard about the subject. When it arrives, AGI will shake life as we know it to its foundations. It could be wonderful, disastrous, or both. It could empower us humans or enslave us. It could save many lives or end many lives. We don’t yet know which of these future outcomes will happen, or even which is most likely. We don’t yet know the mechanisms by which they might occur. But we do know that the advent of artificial general intelligence on earth will be one of the most momentous events to ever take place on it. And we’d better start getting ready. Reading this book is a good start.

Thomas H. Davenport

Distinguished Professor, Babson College

Fellow, MIT Initiative on the Digital Economy

Visiting Professor of Leadership, Brown University

Author of All In on AI, Working with AI, Only Humans Need Apply, The AI Advantage