Responsible AI safeguards

I am fascinated by what we can enable with AI, however, any new technology that is distributed irresponsibly can codify and inflame systems with known bias with potentially devastating impact. Given the broad influence and power that tech companies wield I believe the first things they should consider is both  constraint and a principled base. These principles should be carefully considered and very public by default. Here are some of those underlying principles espoused by Microsoft:

The principles are:

1. Fairness. We will work to develop and deploy facial recognition technology in a manner that strives to treat all people fairly.
2. Transparency. We will document and clearly communicate the capabilities and limitations of facial recognition technology.
3. Accountability. We will encourage and help our customers to deploy facial recognition technology in a manner that ensures an appropriate level of human control for uses that may affect people in consequential ways.
4. Non-discrimination. We will prohibit in our terms of service the use of facial recognition technology to engage in unlawful discrimination.
5. Notice and consent. We will encourage private sector customers to provide notice and secure consent for the deployment of facial recognition technology.
6. Lawful surveillance. We will advocate for safeguards for people’s democratic freedoms in law enforcement surveillance scenarios and will not deploy facial recognition technology in scenarios that we believe will put these freedoms at risk.

More details on these principles can be found here.

Given these principles the Azure Cognitive Services team made public the following principles and standards for Facial recognition technology:

A core priority for the Cognitive Services team is to ensure its AI technology, including facial recognition, is developed and used responsibly. While we have adopted six essential principles to guide our work in AI more broadly, we recognized early on that the unique risks and opportunities posed by facial recognition technology necessitate its own set of guiding principles…

Effective today, new customers need to apply for access to use facial recognition operations in Azure Face API, Computer Vision, and Video Indexer…

Facial detection capabilities (including detecting blur, exposure, glasses, head pose, landmarks, noise, occlusion, and facial bounding box) will remain generally available and do not require an application.

I think this is a very compelling set of necessary promises in a field that is being rapidly deployed to just about every possible corner of our existing technology and cloud services. For me this is about defining reassuring limits, even before our laws catch up to our power, and giving customers real accountability and fairness baked into every AI project.

Disclaimer: While I work for Microsoft these opinions are my own, I have a personal interest in AI, but I am not connected with the teams mentioned as of the date of this writing.