AI Science and Engineering: A New Field (2022) (www.computer.org)

In a 2022 position paper the author proposes "AI Science and Engineering" (AISE) as a distinct, interdisciplinary discipline that unifies foundational science and engineering practice to drive the next-generation AI era. Rather than treating AI purely as an application toolkit, AISE frames a comprehensive body of knowledge—spanning AI fundamentals (neuroscience, evolution, biology, societal awareness), AI technologies (representation, learning, probabilistic programming, robotics, perception, conversation), and AI foundations (philosophy, cognitive science). It also lays out practical engineering layers: AI techniques (chips, models, systems), AI system engineering (requirements, orchestration, benchmarking) and AI management/governance (project management, risk, privacy, ethics). The paper argues this structure should shape AI professions and education for Industry 4.0 and the intelligent digital era. Technically, the paper catalogs eight paradigm shifts that explain AI’s evolution—from general to specialized aims, closed to open domains, hypothesis-driven to hypothesis-free settings (non‑IID data, unknown states), shallow to deep designs (lightweight to deep neural methods like deep Q-learning), individual to hybrid approaches, and small- to large‑scale systems—while highlighting rising complexity and uncertainty as central research challenges. Implications include a push for metasynthetic intelligence (hybridizing diverse intelligences), new benchmarks for real-world, context-aware systems, and curricular/professional reforms to produce engineers who can blend theory, scalable systems, and governance.