This guest post was written by Sunil Soares, Founder and CEO of YDC – AI Governance. He previously founded and led Information Asset, a data management company. Sunil brings a deeply researched perspective to AI governance – author of 13 books that have shaped the way businesses approach data and AI at scale.
The YDC team developed the AI Governance prototype in Atlan. We reused the existing operating model with assets and added custom attributes and relationships.
AI use cases
As discussed in an earlier blog, a digital twin can be a virtual replica of a specific patient that reflects the patient’s unique genetic makeup, or a simulated three-dimensional model that exhibits the characteristics of the patient’s heart. Digital twins can be used to speed up clinical trials and reduce costs in the life sciences industry. The YDC team implemented the overview Digital Twins for Clinical Trials An Artificial Intelligence Use Case in Atlanta.
AI Risk Assessment
We have conducted an AI risk assessment for the use case with Atlan. Digital twins have the potential to introduce bias risks based on algorithms and underlying datasets. We have documented the assessment of risk of bias and mapping to related regulations in Atlan.
We’ve also documented Atlan’s privacy risks.
At Atlan, we have documented other dimensions of AI risk including reliability, accountability, explainability and security. For the sake of brevity, I did not include the screenshots here.
This use case would probably be classified as high risk based on the category of medical devices under Article 6 of the EU AI Act.
AI Risk Assessment Workflows
We have configured the AI Risk Assessment workflow in Atlan to route the AI Risk Assessment to the appropriate parties for approval.
The screenshot below shows the AI Risk Assessment in Approved status based on approval by the Operational Risk Management Committee (ORMC) and the AI Governance Council.
AI shadow control to process metadata from ServiceNow CMDB and face-hugging YDC_AIGOV agents to highlight COTS applications with embedded AI
In an earlier blog, I discussed Shadow AI Governance and YDC_AIGOV agents. As part of the current exercise, we put metadata related to commercial applications (COTS) into Atlan. This information includes metadata such as application name, privacy policy URL, data specifically excluded from AI training, embedded AI, and the option to opt out.
The screenshot below shows Atlan before running the integration with YDC_AIGOV agents. The catalog contains only one AI use case (digital twins for clinical trials) and one application (Google product services).
After launching the integration with the Atlan API, Atlan includes a wider list of applications including Actimize Xceed including metadata in the right panel.
Conditional logic with Atlan API to automatically create AI use cases and AI risk assessment objects
We implemented conditional logic in the Atlan API to automatically create AI use cases only for applications with built-in AI. In this case, we created an AI use case object in Atlan for Actimize Xceed because Embedded AI = “Yes”.
We also implemented conditional logic in the Atlan API to automatically create AI risk assessment objects where data specifically excluded for AI training = “No”. Obviously, this logic is configurable.
This is the basic configuration of AI Governance in Atlan, with more to come!
This post was originally published on Your Data Connect. Read the original article here.