Abstract
This paper discusses the emergence of artificial intelligence (AI) as a new field in engineering. AI first emerged in the 1950s and over the next few decades experienced both advances and obstacles. However, we are now in a new era of AI. The history of AI has been an acceleration from object intelligence (e.g., on symbol, behavior, and agent) to system intelligence (e.g., human, nature, and society), and from individual intelligence (e.g., learning intelligence) to metasynthetic intelligence (hybridizing and synthesizing intelligences). Building on tumultuous AI evolutions, this new-generation AI is accelerating its pace of innovating, differentiating, transforming, and reshaping the world. This paper explores the fundamental questions to be addressed in AI Science and Engineering (AISE), the body of knowledge required, and the professional profiles and educational structures needed to drive the intelligent digital era and Industry 4.0.
Introduction
Since its inception in the 1950s, Artificial Intelligence (AI) has been primarily viewed as a sub-field of computer science. However, as AI technologies increasingly permeate every aspect of human life, economy, and society, this classification is becoming insufficient. The history of AI has been an acceleration from object intelligence—focused on symbols, behaviors, and isolated agents—to system intelligence, which integrates human, natural, and societal dimensions.
Today, we find ourselves in a "New Age" of AI (post-2000), where the focus is shifting toward "Open Complex Intelligent Systems" (OCISs). These systems are not merely collections of algorithms but holistic entities that must operate in open, uncertain environments. To manage this complexity, we must move beyond treating AI as a toolkit and establish it as a formal discipline: AI Science and Engineering (AISE).
The Evolution of AI: From Old Age to New Age
The transition from the "Old Age" of AI (pre-2000) to the "New Age" is marked by profound shifts in philosophy, aims, and methodologies. In the earlier era, the primary goal was to simulate how humans think through mental models and rational agents. In contrast, the current era seeks to develop brain-like cognitive systems and ubiquitous "X-intelligences"—spanning data, networks, and organizational structures.
As shown in Table 1, the foundations of AI have expanded from reductionist logic and probability to include holism, systematism, and a broader range of social sciences like sociology and ethics. The design paradigm has similarly shifted from hypothesis-driven, symbolic approaches to data-driven discovery that incorporates emotion, intention, and feedback.
The AISE Framework: Bridging Science and Engineering
The proposed AI Science and Engineering (AISE) discipline is structured to bridge the gap between theoretical inquiry and industrial application. It can be conceptualized as a dual-faceted framework where "AI Science" provides the theoretical bedrock and "AI Engineering" provides the practical methodologies for implementation and governance.
As illustrated in Figure 1, the AISE body of knowledge is organized into six core pillars:
- AI Fundamentals: Exploring brain and neuroscience, evolution theories, and complex system theories.
- AI Foundations: Drawing from philosophy, cognitive science, mathematics, and sociology.
- AI Technologies: Developing capabilities in analytics, learning, robotics, and intelligent systems.
- AI Engineering Techniques: Focusing on hardware (chips/systems), algorithms, and manufacturing.
- AI System Engineering: Establishing standards for specification, analysis, design, and benchmarking.
- AI Management & Governance: Addressing the critical needs for project management, ethics, privacy, and risk compliance.
Addressing the Professional and Educational Gap
A significant challenge in the current AI landscape is the lack of standardized professional training. Much of what is currently labeled as "AI education" is often criticized as "repackaged data science," focusing heavily on algorithms while neglecting system-level engineering and governance.
To produce qualified AI professionals, the paper advocates for a hierarchical curriculum that spans from primary education to doctoral research. This curriculum must move beyond teaching isolated tools to fostering a deep understanding of "Actionable Intelligence"—intelligence that is not only accurate but also secure, trustworthy, and impactful in real-world business and social contexts.
Future Directions: Managing Complexity and Uncertainty
The frontier of AISE lies in managing "Open Complex Intelligent Systems." Unlike closed-world laboratory models, these systems must adapt to noisy, non-IID (non-independent and identically distributed) data and human-machine cooperation. Future work in the field must prioritize:
- AI Orchestration: Managing AI across hardware, memory, and cloud infrastructures.
- AI Governance: Developing robust KPI metrics to measure the ethical safety and business impact of AI tasks.
- AI Factories: Creating standardized software environments for the rapid construction of reliable AI systems.
Conclusion
AI Science and Engineering is a young but essential discipline. By formalizing the body of knowledge and establishing rigorous standards for education and professional practice, we can move from building isolated "smart" tools to engineering resilient, ethical, and systemic intelligence. This transition is imperative to fully realize the potential of the intelligent digital era and drive the next wave of global innovation.
Summary updated