The SAS Trustworthy AI Life Cycle outlines steps for evaluating and deploying a more trustworthy AI system. It aims to make the U.S. National Institute of Standards and Technology's (NIST) recommendations, standards, and best practices for AI risk management easier to adopt and follow.
The life cycle helps organizational stakeholders specify individual roles and expectations, gather required documentation, and outline factors for consideration. As a result, teams can produce documentation to support the assertion that the organization has done its due diligence to provide evidence that the model is fair and their processes do not cause harm.
The SAS Trustworthy AI Life Cycle is a stepwise process organized into sections. Many tasks in the life cycle reflect guidelines put forth by the NIST AI Risk Management Framework Playbook as it appeared in October 2023.
The Trustworthy AI Life Cycle is designed for teams collaborating to create, test, and implement models. Using the document effectively requires identifying six personas. In some organizations, these personas may be six different individuals—even six different teams. At other organizations, multiple stakeholders may serve several of these roles Begin by identifying the organizational stakeholders who will carry certain responsibilities for the model project. Though titles can differ from one organization to another, the process requires input from the following roles:
- model owner who is the decision-maker
- data scientist who develops the model
- engineer who deploys the model
- risk owner who identifies, manages, and documents risks associated with the model
- data engineer who selects and prepares data for the model
- domain expert who can address business and domain-specific questions.
Note
Any other organizational roles that are associated with modeling could also provide valuable contributions.
Users should be generally familiar with these model management steps:
- data processing
- model creation
- testing
- documentation
- risk management
- deployment
- monitoring
Using the SAS Trustworthy AI Life Cycle requires no additional resources.
Maintainers recommend forking the project into a shared repository so the project team can collaborate on completing it, adding their notes and responses directly to various sections. This way, teams have an auditable record of the document's changes and can more easily share their work with stakeholders and other interested parties.
Teams can also automate their use of the life cycle with a platform or tool of their choosing. For example, a version of the SAS Trustworthy AI Life Cycle is available as a sample workflow that is compatible with SAS® Workflow Manager. This workflow can help users orchestrate trustworthy model creation and management in SAS® Model Manager.
Note
Although the Trustworthy AI Life Cycle sample workflow does not mirror this life cycle document in every respect, improvements to this document can result in enhancement requests for the Trustworthy AI Life Cycle sample workflow.
Maintainers are accepting patches and contributions to this project. Please read CONTRIBUTING.md in the project's repository for details about submitting contributions.
Materials in this repository are licensed under the Apache 2.0 License.
Materials in the /website/docs folder are licensed under the Creative Commons Attribution 4.0 International license.
For more information, see the license in the /website/docs folder.