RAI Construction with Reuse

Summary: It is highly desirable and valuable to ethically reuse the AI artifacts across different applications.

Type of pattern: Process pattern

Type of objective: Trustworthiness

Target users: Developers

Impacted stakeholders: Testers

Lifecycle stages: Implementation

Relevant AI ethics principles: Human, societal and environmental wellbeing, human-centered values, fairness, privacy protection and security, reliability and safety, transparency and explainability, contestability, accountability

Context: Building AI systems from scratch can be very complex and time consuming. Larger companies usually have significant investments in AI and access to large volumes of data, allowing them to compete effectively in the market. In contrast, smaller companies may have only a small team of data scientists, making it challenging for them to compete with larger companies.

Problem: How can we build AI systems cost-effectively?

Solution: It is highly desirable and valuable to reuse AI assets, including AI components and AI pipeline artifacts, across different applications. Construction with reuse refers to the development of RAI systems using existing AI assets, such as those found from an organizational repository or an open-source platform. A marketplace can be established to facilitate the trading of reusable AI assets, including component code, models, and datasets. Blockchain technology can be utilized to create an immutable and transparent marketplace, allowing the auction-based trading of AI assets and material assets, such as cloud resources.


  • Increased development efficiency: Using existing AI assets can significantly accelerate the development process. Reusing the assets that have been previously tested can improve the overall quality of the AI system.
  • Faster time to market: By using existing AI assets, companies can save time and bring their AI systems to market faster.


  • Limited by the quality of AI assets: The quality of reused AI assets may not meet the ethical requirements or have potential RAI risks.
  • Limited customization: The reuse AI assess may limit the degree of customization for a specific application, as the assets may not perfectly align with the requirements.
  • Increased communication cost: Reusing AI assets from external may require additional effort and cost for communication.

Related patterns:

  • Verifiable RAI credential: To ensure the ethical quality, RAI credentials can be bound with the AI assets or developers, which can also be supported by blockchain platforms.

Known uses:

  • pytorch2keras is a model migration tool coverting PyTorch to Keras models.
  • Lewis and Ozkaya highlight glue code needs to be generated for integrating AI components with different systems.
  • SAIaaS is a blockchain-based marketplace for trading AI artifacts and running AI tasks.