As more and more organizations scale their use of AI, they’re challenged with mitigating the associated risks and building genuine trust in AI decision-making. When it comes to trustworthy AI, we believe that consumers, clients and all stakeholders need to know how AI impacts their day-to-day lives, organizations, and work. At IBM, we are working to accelerate the path toward AI transparency. That’s why IBM announced plans for new IBM Watson capabilities and IBM Services designed to help organizations manage AI and build trust throughout its entire lifecycle, from preparing and building AI models to deploying and managing them across the lifecycle.

AI FactSheets and Cloud Pak for Data offer new AI governance tools

Assembling documentation about an AI model’s important features, such as its purpose, performance, datasets, characteristics, and more, can help drive trust in the technology. This is why in 2018, IBM Research proposed the concept of AI FactSheets.

Today, IBM is announcing plans to commercialize key automated documentation capabilities from IBM Research’s AI FactSheets methodology into Watson Studio in Cloud Pak for Data throughout 2021. The capabilities will provide businesses with tools to define how their AI should be used, capture key facts automatically on a model’s performance, and generate reports, or FactSheets, designed to support their internal and external transparency and compliance with regulations. AI FactSheets will offer:

• Policy creation – FactSheet template: Defines what information is collected about models and tracked through the FactSheet, the model facts, such as how an AI service was created, tested, trained, deployed, and evaluated, data use, what regulations or company policies an organization is accounting for, who can use the model and for what purpose, and how it should operate.
• Automated data capture – model facts: Helps to continuously and automatically capture the model facts set forth by the FactSheet template, across the entirety of the AI lifecycle.
• Automated reporting – FactSheet: Provides a FactSheet, compiled in a sharable, comprehensive location that offers knowledge about the AI model in multiple formats, depending on the needs and preferences of different team members and external audiences. The FactSheet tracks the facts as the model is built, updated, and running in production, providing up-to-date insights.

IBM’s Scaling AI Not Risks: Removing Trust as a Barrier to AI Adoption survey, commissioned from Morning Consult, revealed that 68% of AI professionals at larger companies surveyed believe their teams spend a medium or large amount of time documenting their data and AI models for internal and external compliance and reporting, and 84% of AI professionals agree that consumers are more likely to choose services from a company that offers transparency and an ethical framework on how its data and AI models are built, managed and used.

Like nutrition labels for foods or information sheets for appliances, factsheets for AI services would provide information about the product’s important characteristics, helping as organizations streamline their compliance and reporting processes, furthering efforts to build consumer and enterprise trust in AI services.

The AI FactSheet tools are intended to complement IBM Cloud Pak for Data, which recently added new capabilities to provide a comprehensive foundation for AI that can run on any cloud, including a more intuitive role-based user interface, enhanced governance and security, and federated learning to enable model training on distributed data sets while assuring data privacy and security.

New IBM services can help clients build trustworthy AI

Trustworthy AI is about having a holistic approach to AI governance that brings together tools, solutions, practices, and people to govern AI responsibly across its lifecycle. That’s why IBM is launching IBM Services for AI at Scale, a new service offering that provides a framework, methodology, and underlying technology to guide organizations on their AI journey. The service offering is designed to support clients through workstreams such as:

•  AI Ethics assessment: Comprehensive assessments from identifying potential vulnerabilities to understanding potential client ethics maturity and risk.
• Trust-focused AI design: Co-creating the frameworks and principles that garner top-down support and sponsorship, setting clear AI principles for the organization.
• Enterprise-wide innovation: Developing AI and ML models that are ethical, actionable, reusable and scalable.

What it means for clients

Let’s think about what these advancements could mean for a hypothetical major telecommunication company. Today, this company might spend considerable time and resources documenting their deployed AI models for reports to regulators and other interested parties. This documentation might typically be done in an ad hoc, manual and incomplete manner, reducing the effectiveness of the process. By using the proposed AI FactSheets capabilities anticipated for release in Watson in 2021 and insights gained from working with the IBM Services for AI at Scale team, the company could specify precisely which information to collect from their AI models and automatically produce documentation to help them comply with their regulatory obligations. IBM plans to roll out the AI FactSheets throughout 2021. This is designed to help ensure that the company is prepared for their compliance requirements.

To deploy and scale AI, organizations must be able to trust their models and their business outcomes across the entire AI lifecycle. Through our plans to commercialize the capabilities that IBM Research has conceptualized on AI FactSheets into Watson, and launching these new services, we will be providing our clients with a comprehensive portfolio of AI governance solutions that can help increase transparency, manage risk and build greater trust in AI.