AI Principles
We are committed to ensuring that we use AI responsibly, including in compliance with governing laws, privacy and third party rights. As a business we have AI policies and procedures in place governing how AI will be used in our business.
Guidelines for AI-Driven Features in Our Product
The following outlines the position we, as a business, have adopted regarding the use of AI across all our products:
- Transparency: Any feature leveraging AI must clearly communicate to customers that AI is involved. Customers should have the choice to use the feature or decide not to use.
- User-Facing Features: AI-driven features that directly affect end users, such as learners or managers, will be disabled by default. Admins must actively enable these features to ensure deliberate adoption and oversight by each customer.
- Admin-Facing Features: For AI-powered features intended for use by admins, we will assess and provide choice and configurability, guided by user feedback and our internal AI governance process. Where AI is used it will be clearly stated and will be the Admins choice whether to use.
- Documentation: We will maintain transparency about the technology and data involved in our features, clearly outlining what is and isn’t being used. For AI-specific features, currently this information will be provided through Zendesk articles to ensure accessibility and clarity for our users.
Internal Governance
At Kallidus, we conduct internal governance meetings to evaluate all proposed uses of AI. These meetings bring together key stakeholders, including members of the Senior Leadership Team, our AI Technical Lead, a Product Manager, and consultation with our Compliance and Legal teams. This ensures a comprehensive review of both technical feasibility and legal compliance, particularly regarding contractual terms and data usage.
The committee is responsible for making go/no-go decisions on proposed initiatives. In some cases, they may introduce specific restrictions to ensure alignment with ethical standards, business objectives, and regulatory requirements.
AI Technology
We currently use a mixture of Azure OpenAI GPT-4o models, and proprietary ML models, in our AI-driven product features.
Bias and Fairness Audits
As part of our internal AI governance approval process, we include specific questions to identify who might be affected by the AI-driven feature, in what ways, and how potential bias or discrimination can be mitigated. Additionally, we allow for the role of a "human in the loop" to ensure oversight, and for opt-in or opt-out options for using the AI-driven feature.
For more information on specific AI-driven features you can review these articles: