This book offers readers a fresh perspective on intelligence, its alignment with human society, and the role AI plays in shaping our collective future. By exploring AI across three perspectives, external AI, internal AI, and societal AI, the book addresses the evolving challenges each stage presents.
Reconceiving AI: The World as an Apple or a Blue Orange explores the diverse nature of intelligence and addresses the alignment problem by questioning the types of intelligence we seek in AI. By examining neurotypical human emotions such as love, empathy, and remorse, and intrinsic moral imperatives, it proposes a new form of intelligence, neurodivergent yet harmoniously aligned with human society. The book draws on Cézanne’s still-life paintings to illustrate the relational and substantive nature of objects within his world models and uses Paul Éluard’s poem The World Is a Blue Orange to explore how imagination and creativity can deepen our understanding of these models. Éluard’s surreal imagery challenges our conventional understanding of reality, prompting us to consider whether AI could adopt similarly creative, non-linear ways of perceiving and representing the world. The exploration continues with moral and ethical dilemmas found in Eric Rohmer’s Six Moral Tales and Krzysztof Kieślowski’s Dekalog films, shedding light on how these principles should be reflected in the AI systems of the future. Finally, the book offers an in-depth analysis of generative AI technologies, particularly large language models and multimodal systems, evaluating their scientific limitations and their potential to advance AI toward a truly intelligent system.
This book equips readers with the tools to understand AI’s potential impact on society and its ethical and scientific challenges. It will be useful to developers, researchers, students, and general readers. The book provides an accessible, thought-provoking exploration of how AI can evolve and what we should expect from it.