『How to Create a Mind』要約

In How to Create a Mind, inventor and futurist Ray Kurzweil argues that reverse-engineering the human brain to build an artificial general intelligence is not only feasible but imminent. Central to his thesis is the Pattern Recognition Theory of Mind, which holds that the neocortex is essentially a hierarchical system of roughly 300 million pattern recognizers. Each module learns a pattern, flags it when encountered again, and passes the signal up a hierarchy of increasing abstraction—from edges and shapes to faces, words, and concepts. Kurzweil illustrates this with everyday examples like recognizing a friend's face or understanding a sentence, and he ties the model to the known columnar architecture of the cortex. The book draws heavily on Kurzweil's own career, including his work on speech recognition, reading technology, and music synthesis, and it revisits his law of accelerating returns and the concept of the singularity. He describes projects such as the Blue Brain Project and IBM's Watson to show how brain-inspired computing is progressing, and he discusses the role of hidden Markov models and hierarchical systems in both biological and machine intelligence. A key argument is that the brain's algorithms are relatively simple and repeated massively, making them learnable and replicable. Kurzweil also addresses objections from critics like Douglas Hofstadter and explores the ethical and existential implications of mind uploading and superintelligence. He predicts that by the 2020s and 2030s, we will have the tools to create a synthetic neocortex, leading to machines that rival and then surpass human intelligence. The book blends neuroscience, computer science, and futurism into a single, ambitious claim: that creating a mind is an engineering problem we are on track to solve.