Expert Systems in Design

1. Overview

Expert Systems in Design are a class of artificial intelligence (AI) applications that capture and leverage the knowledge of human experts to solve complex design problems. These systems act as intelligent assistants to designers, providing specialized knowledge, automating routine tasks, and offering design recommendations. The core idea is to codify the heuristics, rules of thumb, and experiential knowledge of seasoned designers into a machine-readable format, making it accessible to a wider range of users and preserving valuable expertise. By doing so, expert systems can significantly enhance design productivity, improve design quality, and facilitate the exploration of a larger design space. The origin of expert systems can be traced back to the early days of AI research in the 1960s and 1970s, with pioneering systems like DENDRAL and MYCIN demonstrating the feasibility of capturing and applying expert knowledge. The application of these concepts to design domains followed soon after, driven by the need to tackle increasingly complex engineering and architectural challenges.

2. Core Principles

1. Separation of Knowledge and Inference: A fundamental principle of expert systems is the clear distinction between the knowledge base, which contains the domain-specific facts and heuristics, and the inference engine, which is a general-purpose reasoning mechanism that applies the knowledge to solve problems. This separation allows for easier maintenance and modification of the knowledge base without altering the underlying reasoning system.

2. Explicit Knowledge Representation: The knowledge of human experts is made explicit and is represented in a formal, symbolic language. Common knowledge representation formalisms include production rules (IF-THEN statements), frames, and semantic networks. This explicit representation makes the knowledge inspectable, understandable, and transferable.

3. Heuristic-Based Reasoning: Expert systems often rely on heuristics, or rules of thumb, which are plausible but not always guaranteed to lead to the optimal solution. These heuristics capture the experiential knowledge and intuition of human experts, enabling the system to solve problems that are ill-structured or have incomplete information.

4. Explanation and Transparency: A key feature of expert systems is their ability to explain their reasoning process. By tracing the chain of inferences that led to a particular conclusion, the system can provide a justification for its recommendations. This transparency is crucial for building trust with users and for debugging the system’s knowledge base.

3. Key Practices

1. Knowledge Engineering: This is the process of eliciting, analyzing, and codifying the knowledge of human experts. It involves a series of interviews and observations with domain experts to understand their problem-solving methods and to translate their expertise into a formal representation that can be used by the expert system.

2. Rule-Based Programming: A common practice in developing expert systems is to use rule-based programming, where knowledge is represented as a set of IF-THEN rules. These rules are then processed by an inference engine to derive conclusions. For example, a rule in a structural design expert system might be: “IF the beam span is greater than 20 feet, THEN use a steel I-beam.”

3. Frame-Based Representation: In this approach, knowledge is organized into frames, which are data structures that represent stereotyped situations or objects. Each frame has a set of slots that can be filled with specific values. For example, a “beam” frame might have slots for material, cross-section, and length. Frames can be organized into hierarchies, allowing for inheritance of properties.

4. Case-Based Reasoning (CBR): Instead of relying on general rules, CBR systems solve new problems by retrieving and adapting solutions from similar problems that have been solved in the past. In a design context, a CBR system might suggest a design for a new building by adapting the design of a similar building from its case base.

5. Constraint Checking: Expert systems can be used to automatically check whether a design satisfies a set of predefined constraints. These constraints can be related to physical laws, building codes, manufacturing limitations, or project requirements. For example, an expert system for circuit design could check for violations of electrical rules.

4. Application Context

• Best Used For:
  • Conceptual Design: Assisting designers in the early stages of design by providing suggestions, evaluating alternatives, and checking for feasibility.
  • Configuration and Layout: Automating the process of configuring complex systems, such as computer networks or office layouts, based on a set of requirements and constraints.
  • Design for Manufacturing (DFM): Providing feedback to designers on the manufacturability of their designs, helping to reduce production costs and improve quality.
  • Troubleshooting and Diagnostics: Assisting engineers in diagnosing faults in complex systems by reasoning through a set of symptoms and test results.
  • Knowledge Capture and Training: Preserving the knowledge of experienced designers and using it to train new designers.
• Not Suitable For:
  • Highly Creative or Innovative Design: Expert systems are best suited for well-defined problems where the design knowledge can be easily codified. They are less effective in situations that require a high degree of creativity or out-of-the-box thinking.
  • Poorly Understood Domains: If the design knowledge is not well understood or is constantly changing, it can be difficult to build and maintain an effective expert system.
• Scale: Individual/Team/Department/Organization
• Domains:
  • Engineering: (e.g., aerospace, automotive, civil, and mechanical engineering) for tasks such as structural analysis, circuit design, and process planning.
  • Architecture: for tasks such as building layout, code compliance, and energy analysis.
  • Software Engineering: for tasks such as program synthesis, debugging, and maintenance.
  • Manufacturing: for tasks such as process planning, scheduling, and quality control.

5. Implementation

• Prerequisites:
  • Access to Domain Expertise: The most critical prerequisite is access to one or more human experts who are willing and able to articulate their knowledge.
  • Knowledge Engineering Skills: The development team needs individuals with skills in knowledge engineering to elicit, analyze, and represent the expert’s knowledge.
  • Appropriate Tools and Technology: This includes expert system shells, which provide a pre-built inference engine and knowledge base structure, or programming languages like Prolog or Lisp that are well-suited for symbolic reasoning.
• Getting Started:
  1. Problem Selection: Identify a suitable design problem that is well-scoped and for which there is a clear need for an expert system.
  2. Knowledge Acquisition: Conduct a series of interviews and observations with domain experts to gather the necessary knowledge.
  3. Knowledge Representation: Choose an appropriate formalism (e.g., rules, frames) to represent the acquired knowledge.
  4. Prototyping and Iteration: Build a small-scale prototype of the expert system and then iteratively refine and expand it based on feedback from experts and users.
  5. Testing and Validation: Thoroughly test the system to ensure that it provides accurate and reliable advice.
• Common Challenges:
  • The Knowledge Acquisition Bottleneck: Eliciting and codifying knowledge from human experts is often the most difficult and time-consuming part of building an expert system.
  • Handling Uncertainty and Incomplete Information: Design problems often involve uncertainty and incomplete information. The expert system must be able to reason with this uncertainty to provide useful advice.
  • Maintaining and Updating the Knowledge Base: As the design domain evolves, the knowledge base must be updated to reflect the latest knowledge and best practices.
• Success Factors:
  • Strong Management Support: Building an expert system can be a significant undertaking, and it requires strong support from management to succeed.
  • User Involvement: Involving end-users throughout the development process is crucial to ensure that the system meets their needs and is easy to use.
  • Realistic Expectations: It is important to have realistic expectations about what the expert system can and cannot do. Expert systems are tools to assist designers, not to replace them.

6. Evidence & Impact

• Notable Adopters:
  • Digital Equipment Corporation (DEC): DEC’s XCON (eXpert CONfigurer) is one of the most famous and successful expert systems. It was used to configure VAX computer systems, a complex task that was previously done by human experts. XCON significantly reduced the error rate and the time required for configuration.
  • Boeing: Boeing has used expert systems for a variety of design and manufacturing tasks, including the design of electrical wiring harnesses and the diagnosis of manufacturing defects.
  • Schlumberger: This oilfield services company developed the Dipmeter Advisor, an expert system that interprets data from oil well logs to help geologists locate oil and gas reserves.
  • General Electric: GE developed an expert system called DELTA (Diesel-Electric Locomotive Troubleshooting Aid) to assist maintenance personnel in diagnosing and repairing diesel-electric locomotives.
  • American Express: American Express developed the Authorizer’s Assistant, an expert system to help credit authorizers decide whether to approve unusual credit requests.
• Documented Outcomes:
  • Reduced Costs: By automating design tasks and reducing errors, expert systems have been shown to significantly reduce design and manufacturing costs.
  • Improved Quality: Expert systems can help to improve design quality by ensuring that designs are complete, consistent, and compliant with all relevant constraints.
  • Increased Productivity: By automating routine tasks and providing intelligent assistance, expert systems can significantly increase the productivity of designers and engineers.
  • Knowledge Preservation: Expert systems provide a mechanism for capturing and preserving the valuable knowledge of experienced designers, which might otherwise be lost when they retire or leave the company.
• Research Support:
  • Numerous studies have been published in academic journals and conference proceedings documenting the development and application of expert systems in various design domains. These studies have provided evidence for the effectiveness of expert systems in improving design processes and outcomes.
  • The development of expert systems has also been a major driver of research in artificial intelligence, particularly in the areas of knowledge representation, reasoning, and machine learning.

7. Cognitive Era Considerations

• Cognitive Augmentation Potential:
  • Machine Learning for Knowledge Acquisition: Modern machine learning techniques can be used to automatically learn rules and relationships from design data, overcoming the knowledge acquisition bottleneck that has traditionally plagued expert systems.
  • Generative AI for Design Exploration: Generative models can be used to create a vast number of design alternatives, which can then be evaluated and refined by the expert system, expanding the creative possibilities for designers.
  • Natural Language Processing for User Interaction: Large language models (LLMs) can provide a more natural and intuitive conversational interface for expert systems, making them easier to use for non-experts.
• Human-Machine Balance:
  • Human Strengths: Creativity, intuition, ethical judgment, and the ability to understand the broader context of a design problem remain uniquely human strengths. Designers will continue to play a critical role in setting design goals, evaluating trade-offs, and making final design decisions.
  • Machine Strengths: Expert systems excel at tasks that require speed, accuracy, and the ability to process large amounts of information. They can handle the tedious and repetitive aspects of design, freeing up human designers to focus on more creative and strategic work.
  • Collaborative Partnership: The future of design is likely to be a collaborative partnership between humans and machines, where expert systems act as intelligent assistants that augment and amplify the capabilities of human designers.
• Evolution Outlook:
  • Hybrid AI Systems: The distinction between rule-based expert systems and data-driven machine learning systems is likely to blur, with the emergence of hybrid systems that combine the strengths of both approaches.
  • Cloud-Based Design Services: Expert systems will increasingly be delivered as cloud-based services, making them more accessible to a wider range of users and enabling new forms of collaboration.
  • Integration with Digital Twins: Expert systems will be integrated with digital twins of physical assets, allowing for real-time monitoring, analysis, and optimization of design performance.

8. Commons Alignment Assessment (v2.0)

This assessment evaluates the pattern based on the Commons OS v2.0 framework, which focuses on the pattern’s ability to enable resilient collective value creation.

1. Stakeholder Architecture: Expert Systems in Design primarily define stakeholders as the designers using the system, the domain experts providing the knowledge, and the organizations deploying it. The rights and responsibilities are heavily skewed towards the organization, which typically owns the intellectual property of the captured knowledge and the resulting designs. This model fails to adequately recognize the rights of the knowledge creators (the experts) or the end-users of the designs.

2. Value Creation Capability: The pattern excels at creating economic value by enhancing design productivity, reducing errors, and optimizing complex designs. However, its capacity for creating broader social, ecological, or knowledge value is limited by its proprietary nature. While it preserves expert knowledge, this knowledge is often siloed within the organization rather than contributing to a wider collective intelligence.

3. Resilience & Adaptability: The resilience of an expert system is contingent on the continuous maintenance and updating of its knowledge base, which requires a significant, long-term investment from the host organization. The system itself is not inherently adaptive to change; it is a snapshot of expertise at a particular time. Its ability to thrive on change is therefore limited and dependent on external intervention.

4. Ownership Architecture: Ownership is traditionally defined in terms of monetary equity and intellectual property rights, which are centralized in the organization that develops the system. The pattern lacks a framework for distributing rights and responsibilities among other stakeholders, such as the experts who contribute their knowledge or the community that uses the resulting products. This represents a significant gap in its alignment with a commons-based approach.

5. Design for Autonomy: Expert systems are a form of artificial intelligence and are highly compatible with distributed systems and DAOs, requiring low coordination overhead once the knowledge base is established. This makes them well-suited for integration into autonomous and semi-autonomous design processes. This is a strong point of alignment with the principles of the Commons OS v2.0 framework.

6. Composability & Interoperability: The pattern is highly composable and can be integrated with other software systems, such as CAD tools, simulation software, and project management platforms. This interoperability allows for the creation of larger, more sophisticated value-creation systems. This is a key enabler for building complex, multi-pattern solutions.

7. Fractal Value Creation: The core logic of separating the knowledge base from the inference engine is a fractal pattern that can be applied at multiple scales. This allows for the development of nested hierarchies of expert systems, from individual design assistants to large-scale, organization-wide knowledge infrastructures. This demonstrates a strong potential for scalable value creation.

Overall Score: 3 (Transitional)

Rationale: Expert Systems in Design have significant potential to contribute to collective value creation but are held back by a legacy ownership and stakeholder model. While they demonstrate strong composability and fractal properties, their full potential as a commons is unrealized due to the centralization of rights and value capture. To become more aligned, the pattern would need to incorporate more equitable models of knowledge ownership and governance.

Opportunities for Improvement:

• Develop a framework for shared ownership of the knowledge base with the contributing experts.
• Implement a governance model that allows for community participation in the evolution of the system.
• Explore the use of open-source expert systems to create a more accessible and collaborative design infrastructure.

9. Resources & References

• Essential Reading:
  • Durkin, J. (1994). Expert Systems: Design and Development. Macmillan.
  • Jackson, P. (1999). Introduction to Expert Systems. Addison-Wesley.
  • Waterman, D. A. (1986). A Guide to Expert Systems. Addison-Wesley.
  • Lucas, P. J. F., & Van Der Gaag, L. C. (1991). Principles of Expert Systems. Addison-Wesley.
• Organizations & Communities:
  • Association for the Advancement of Artificial Intelligence (AAAI): A non-profit scientific society devoted to advancing the scientific understanding of the mechanisms underlying thought and intelligent behavior and their embodiment in machines.
  • AIGA, the professional association for design: AIGA is the oldest and largest professional membership organization for design.
• Tools & Platforms:
  • CLIPS: A public domain software tool for building expert systems.
  • Prolog: A logic programming language that is well-suited for developing expert systems and other AI applications.
  • Lisp: A family of computer programming languages with a long history of use in AI research and development.
• References:
  • Wikipedia contributors. (2023). Expert system. In Wikipedia, The Free Encyclopedia. Retrieved from https://en.wikipedia.org/wiki/Expert_system
  • Ó Catháin, C. (1987). Expert systems and design. Design Studies, 8(2), 58-61. https://doi.org/10.1016/0142-694X(87)90001-9
  • Lucas, P. J. F., & Van Der Gaag, L. C. (1991). Principles of Expert Systems. Addison-Wesley.
  • McDermott, J. (1982). R1: A rule-based configurer of computer systems. Artificial intelligence, 19(1), 39-88.
  • Sriram, D., & Adey, R. (Eds.). (2012). Applications of artificial intelligence in engineering VI. Springer Science & Business Media.