Artificial Intelligence: A Guide for Thinking Humans

As one of the founding figures of AI and a cofounder of MIT’s AI Laboratory, Marvin Minsky shaped early conceptions of machine reasoning, symbolic representation, and cognitive architecture. His influence permeates the field’s history, and Mitchell frequently cites him to illustrate both the achievements and the missteps of early AI optimism. Minsky championed the idea that intelligence could be understood through modular computational mechanisms—which he later called “the society of mind.”

In Artificial Intelligence, Mitchell refers to Minsky as both an inspiration and a cautionary figure. His belief in the possibility of humanlike machine intelligence helped launch decades of research, yet his famous maxim, “‘easy things are hard’” (33), underscores one of the book’s recurring themes: Tasks that seem trivial to humans often prove enormously difficult for AI systems. Mitchell draws on Minsky’s work to highlight the limits of symbolic systems, the challenges of commonsense reasoning, and the pitfalls of trying to codify intuitive human knowledge into rigid computational rules.

In addition, Minsky’s broader philosophical perspective shapes the conversation about understanding and consciousness. He argued that human mental terms like belief, meaning, or self were temporary placeholders for more precise scientific concepts that had yet to be discovered. Mitchell uses this view to emphasize how early assumptions about intelligence still shape contemporary debates, reminding readers that the conceptual foundations of AI remain incomplete and contested.

Andrew Ng is a leading AI researcher, entrepreneur, and educator known for his foundational contributions to deep learning and online education. As a cofounder of Google Brain and former chief scientist at Baidu, Ng has been central to scaling neural-network research and bringing large-scale machine learning into real-world applications. His work exemplifies the data-driven, deep-learning revolution that reshaped AI in the 2010s, enabling breakthroughs in speech recognition, image classification, and NLP.

Mitchell references Ng throughout the book as a representative voice of contemporary machine learning’s strengths and limitations. He is especially prominent in discussions of autonomous vehicles, arguing that pedestrians must eventually learn to behave more predictably around self-driving cars—a view that highlights the practical and philosophical tensions between human behavior and machine reliability. His confidence in deep learning’s scalability contrasts with Mitchell’s more cautious stance, creating a productive dialogue about what present systems can and cannot achieve.

Within the narrative, Ng symbolizes the modern, engineering-focused mindset in AI—optimistic, ambitious, and oriented around empirical progress. His views help readers understand why deep learning is both celebrated and criticized: It delivers remarkable performance yet still confronts significant challenges related to commonsense reasoning, robustness, and societal impact.

Composer and music theorist David Cope is the creator of Experiments in Musical Intelligence (EMI), one of the most controversial and influential early systems for algorithmic music composition. Cope designed EMI to analyze the stylistic “signatures” of classical composers and generate new compositions that mimicked those styles. His work ignited debates about creativity, authorship, and the boundaries between human and machine-generated art—debates at the heart of Mitchell’s exploration of artificial creativity.

In Artificial Intelligence, Mitchell uses Cope’s EMI project as a vivid case study of how AI can appear creative while lacking true understanding or aesthetic judgment. EMI helps her illustrate the distinction between generating patterns and grasping their meaning: Although EMI produced music that even professional musicians occasionally mistook for Mozart or Bach, the system had no internal grasp of melody, emotion, or musical intent. Its compositions were powerful precisely because they exposed the tension between surface patterning and deeper human artistry.

Cope’s own decisions (including his later destruction of EMI’s signature database) further enrich the narrative. His unease with public backlash, questions of artistic authenticity, and concerns about infinite reproducibility amplify Mitchell’s broader theme: AI achievements often raise complex cultural and philosophical questions that exceed technical performance alone. Through Cope, readers encounter the broader stakes of AI creativity—not just what machines can make, but what human meaning-making requires and what the human spirit can bring to the creative process.

One of the foundational figures in computer science and AI is Alan Turing. His work on computation, formal systems, and machine intelligence, especially through the 1950 paper “Computing Machinery and Intelligence,” laid conceptual groundwork for the questions the field still asks today. Mitchell references Turing primarily through the historical legacy of the Turing Test and the broader philosophical inquiries he posed about the nature of thought, imitation, and machine “understanding.”

In the context of Artificial Intelligence, Turing represents both the ambition and the ambiguity inherent in early AI. His imitation-based framing continues to influence how the public and researchers evaluate AI behavior—often leading people to equate convincing performance with genuine understanding, a conflation that Mitchell repeatedly pushes back against. In addition, Turing’s work is central to later debates about general intelligence: His speculative predictions and methodological optimism set expectations that still shape contemporary discourse.

Mitchell uses Turing as a reminder that the field’s deepest tensions (between simulation and understanding, and between behavior and cognition) were present from its inception. His historical presence gives readers a sense of continuity: AI’s present challenges echo questions Turing raised decades ago, underscoring the work the field still faces to resolve its foundational philosophical problems.

Prominent AI researcher Pedro Domingos is best known for his work in machine learning and his book The Master Algorithm. Mitchell cites Domingos as one of the field’s sharpest critics of symbolic AI approaches, particularly in relation to Douglas Lenat’s Cyc project. His memorable line describing Cyc as “‘the most notorious failure in the history of AI’” (250) encapsulates ongoing debates over whether human knowledge can be hand-coded into logical rules.

Domingos’s role in the book is twofold. First, he represents a mainstream machine-learning perspective skeptical of early symbolic ambitions and highly attuned to the need for scalable, data-driven approaches. Second, his observations about AI safety and reliability illustrate the field’s growing awareness of limitations in machine intelligence. Mitchell draws on his often-pithy insights to highlight the brittleness and vulnerability of modern systems.

In the broader narrative, Domingos’s voice anchors the book’s recurring theme that AI progress is uneven and that public hype frequently outstrips technical reality. His critiques help readers understand why certain approaches have fallen out of favor, and why the field continues to wrestle with fundamental questions about how intelligence should be modeled.

Rodney Brooks is a pioneering roboticist and former director of MIT’s Computer Science and AI Laboratory. Known for championing the “embodiment hypothesis,” Brooks argues that intelligence cannot be achieved through abstract computation alone; instead, it must emerge from systems embedded in and interacting with the physical world. Mitchell cites Brooks repeatedly as a counterweight to disembodied notions of AI, including early symbolic systems and later deep-learning models that lack physical grounding.

Within Artificial Intelligence, Brooks’s ideas help frame Mitchell’s exploration of intuition, commonsense reasoning, and the complex interplay between perception and action. Mitchell uses his critiques to illustrate why approaches like Cyc faltered: They attempted to encode humanlike knowledge without the sensorimotor experiences that inform it. Brooks also provides important historical context, bridging early robotics and contemporary debates about general intelligence.

Brooks’s perspective reinforces one of the book’s central conclusions: that the richness of human cognition may be inseparable from the messy, embodied human experience of the world. By highlighting his skepticism toward grandiose AI predictions, Mitchell positions Brooks as a thoughtful realist whose work underscores how challenging it is to replicate human intelligence in machines.