This visitor publish is by Sunil Soares, founder and CEO of YDC – AI Governance. Beforehand, he based and led Data Asset, an information administration agency. Sunil brings a deeply-researched perspective to AI governance—authoring 13 books which have formed how enterprises strategy information and AI at scale.
The YDC crew developed an AI Governance prototype in Atlan. We reused the prevailing working mannequin with property and added customized attributes and relations.
AI Use Instances
As mentioned in an earlier weblog, a digital twin could also be a digital duplicate of a specific affected person that displays the distinctive genetic make-up of the affected person or a simulated three-dimensional mannequin that displays the traits of a affected person’s coronary heart. Digital twins could also be utilized to speed up scientific trials and scale back prices within the life sciences trade. The YDC crew applied an summary of the Digital Twins for Scientific Trials AI Use Case in Atlan.
AI Danger Assessments
We performed an AI Danger Evaluation for the use case with Atlan. Digital twins have the potential to introduce bias dangers based mostly on the algorithms and the underlying information units. We documented the bias danger evaluation and a mapping to the related rules in Atlan.
We additionally documented the privateness dangers in Atlan.
We documented different dimensions of AI danger together with Reliability, Accountability, Explainability and Safety in Atlan. For the sake of brevity, I’ve not included these screenshots right here.
This use case would probably be categorized as Excessive Danger based mostly on the Medical System class of Article 6 of the EU AI Act.
AI Danger Evaluation Workflows
We configured an AI Danger Evaluation workflow in Atlan to route the AI Danger Evaluation to the suitable events for approval.
The screenshot under exhibits the AI Danger Evaluation in Authorised standing based mostly on approvals from the Operational Danger Administration Committee (ORMC) and the AI Governance Council.
Shadow AI Governance to Ingest Metadata from ServiceNow CMDB and YDC_AIGOV Brokers on Hugging Face to Spotlight COTS Apps with Embedded AI
In an earlier weblog, I mentioned Shadow AI Governance and the YDC_AIGOV brokers. As half of the present train, we ingested metadata across the Business-off-the-Shelf (COTS) apps into Atlan. This info consists of metadata corresponding to Software Title, Privateness Coverage URL, Knowledge Particularly Excluded from AI Coaching, Embedded AI and Choose-Out Choice.
The screenshot under exhibits Atlan earlier than operating the mixing with the YDC_AIGOV brokers. The catalog solely comprises one AI Use Case (Digital Twins for Scientific trials) and one utility (Google Product Providers).
After operating the mixing with Atlan API, Atlan comprises a broader checklist of functions together with Actimize Xceed together with metadata in the proper panel.
Conditional Logic with Atlan API to Auto-Create AI Use Case and AI Danger Evaluation Objects
We applied conditional logic within the Atlan API to auto-create AI use circumstances just for functions with embedded AI. On this case, we created an AI use case object in Atlan for Actimize Xceed as a result of Embedded AI = “Sure.”
We additionally applied conditional logic within the Atlan API to auto-create AI Danger Evaluation objects the place Knowledge Particularly Excluded for AI Coaching = “No.” Clearly, this logic is configurable.
This can be a primary AI Governance configuration in Atlan with extra to come back!
This publish was initially revealed on Your Knowledge Join. Learn the unique article right here.
