Guidance for integration and responsible use with Custom Vision
Microsoft wants to help you responsibly develop and deploy solutions that use Azure AI Custom Vision. We're taking a principled approach to upholding personal agency and dignity by considering the AI systems' fairness, reliability and safety, privacy and security, inclusiveness, transparency, and human accountability. These considerations represent our commitment to developing Responsible AI.
General guidance
When you're getting ready to deploy Custom Vision, the following activities help to set you up for success:
Understand what it can do: Fully assess Custom Vision to understand its capabilities and limitations. Understand how it will perform in your particular scenario and context, by thoroughly testing it with real-life conditions and data.
Respect an individual's right to privacy: Only collect data and information from individuals for lawful and justifiable purposes. Only use data and information that you have consent to use for this purpose. For example, evaluate camera locations and positions. Adjust angles and the region of interest, so they don't monitor protected areas, such as bathrooms, or public spaces, such as external streets or mall concourses.
Legal review: Obtain appropriate legal advice to review your solution, particularly if you will use it in sensitive or high-risk applications. Understand what restrictions you might need to work within, and your responsibility to resolve any issues that might come up in the future. Do not provide any legal advice or guidance.
Human in the loop: Keep a human in the loop, and include human oversight as a consistent pattern area to explore. This means ensuring constant human oversight of the AI-powered product or feature, and maintaining the role of humans in decision-making. Ensure that you can have real-time human intervention in the solution to prevent harm. This enables you to manage situations when the AI system does not perform as required.
Security: Ensure your solution is secure, and that it has adequate controls to preserve the integrity of your content and prevent unauthorized access.
Build trust with affected stakeholders: Communicate the expected benefits and potential risks to affected stakeholders. Help people understand why the data is needed and how the use of the data will lead to their benefit. Describe data handling in an understandable way.
Customer feedback loop: Provide a feedback channel that allows users and individuals to report issues with the service after it has been deployed. Monitor and improve the AI-powered product or feature on an ongoing basis. Be ready to implement any feedback and suggestions for improvement. Establish channels to collect questions and concerns from affected stakeholders (people who might be directly or indirectly impacted by the system, including employees, visitors, and the general public). For example:
- Feedback features built into app experiences.
- An easy-to-remember email address for feedback.
- Anonymous feedback boxes placed in semi-private spaces.
- Knowledgeable representatives in the lobby.
Feedback: Seek out feedback from a diverse sampling of the community during the development and evaluation process (for example, historically marginalized groups, people with disabilities, and service workers). For more information, see Community jury.
User Study: Any consent or disclosure recommendations should be framed in a user study. Evaluate the first and continuous-use experience with a representative sample of the community to validate that the design choices lead to effective disclosure. Conduct user research with 10-20 community members (affected stakeholders) to evaluate their comprehension of the information and to determine if their expectations are met.