Artificial intelligence (AI) has become an integral part of numerous industries, revolutionizing how businesses operate and customers interact with services. However, with great power comes great responsibility. As organizations increasingly rely on AI algorithms to make critical decisions, there arises a pressing need for effective AI governance. This article looks at some of the software tools currently available for governing AI projects.
AI Verify
License: Open Source Apache-2.0 license
Deployment options: Self managed.
AI Verify is an AI governance testing framework and software toolkit that validates the performance of AI systems against a set of internationally recognised principles through standardised tests. AI Verify is consistent with international AI governance frameworks such as those from European Union, OECD and Singapore.
It is a single integrated toolkit that operates within an enterprise environment. It can perform technical tests on common supervised learning classification and regression models for most tabular and image datasets. It however does not define AI ethical standards and does not guarantee that any AI system tested will be free from risks or biases or is completely safe.
If you just wish to test AI Verify without having to install it, we have prepared the AI Verify Playground precisely for this reason.
We hope you take it for a spin and provide us feedback.
And to aid your understanding of the software, I have also prepared a demo video to follow along.
Credo AI
Deployment options: Fully managed.
Credo AI positions itself as empowering organisations to create AI with the highest ethical standards by allowing business and technical stakeholders to measure, manage, and monitor AI-introduced risks across data, models, and processes to ensure responsible, auditable, and compliant AI at scale.
With Credo AI customers can register all their AI projects in a common registry named the Use Case Registry. Each AI use case has an AI use case card, which is an overview of the current risk level of the AI solution based on the current state of governance. Businesses can then enforce governance through policy packs. Policy packs are modular sets of policies and controls that are designed to address objective technical risks and process-based compliance.
Watch this session from Susannah Shattuck, Head of Product at Credo AI, who shares best practices from the field and tactical approaches that you can begin using today to streamline your AI Governance.
Modulos AI
Deployment options: Self managed.
Modulos AI is a Zurich based startup providing an end to end platform for data science with a special focus on governance and automated documentation. It has features like policy templates, upstream monitoring, centralized artifact monitoring and reporting, in built into the platform.
An added advantage of this platform is that its available for deployment on-premise or even inside customers virtual private cloud on AWS, Azure and GCP.
Holistic AI
Deployment options: Fully managed & self managed.
Holistic AI provides comprehensive and modular AI GRC capabilities that promise to transform how AI is adopted and used within enterprises. It provides
A command centre suite for executive-level management of AI applications
Provides risk monitoring and compliance reporting.
Provides mitigation recommendations to improve the trustworthiness of the system.
DataRobot
DataRobot is an end to end ML platform which comes with a use case registry, automated model documentation and advanced MLOPS features that include monitoring as well. While the platform is not exclusively designed for AI governance, the governance features have always been a core differentiator of the platform. Compliance documentation for DataRobot can be generated at a click of a button and there is also support for automated compliance documentation for external models which have been created outside of the platform.
IBM Watson OpenScale
Deployment options: Fully managed & self managed.
IBM Watson OpenScale is an end to end model monitoring service which provides plug and play integration with technologies such as AWS Sagemaker which is used to build and deploy models. Watch the demo below to know more about its features.
Conclusion
The AI governance space is still in its nascent stages with most organisation resorting to manual documentation and reporting. With the legal frameworks starting to catch up and enforce compliance, it will become imperative for enterprises to track and document their AI solutions with automation.
Bookmark this blog for updates on AI governance solutions and recent developments in this area.
