Understanding Quantum Workplace’s AI Tools, Data Privacy, and Governance
Last Updated: December 11, 2024
In this article:
Overview
Quantum Workplace uses artificial intelligence (AI), tools in several areas within its Software as a Service (Saas) product. These AI tools are powered by the Microsoft Open AI platform, an installed and fully contained service within the Quantum Workplace Azure environment.
The Open AI platform does not receive any inputs from data entered into the Quantum Workplace (Microsoft Open AI) system, nor does it provide outputs that the Quantum Workplace system would display. Read more about Microsoft Open AI.
Quantum Workplace's AI Tools
- Ask AI: Processes quantitative and qualitative employee opinion data
- Writing Assistant: Processes free-form text to refine content and make
suggestions for improvement - Goal Assistant: Aids users in creating personal, team, and company goals by
processing goal descriptions - Summary Assistant: This tool processes free-form text entered into open-ended
survey questions to generate concise summaries - Discussion Assistant: Assists users in providing feedback by processing their
responses - Action Assistant: Processes quantitative data from survey results, user input in
the form of discussion notes, the organization's industry, and the user's job title
The data processed by these tools often depends on user inputs and/or filters/slices applied by a user. For example, the user can filter a particular employee demographic to narrow the scope of the Summary Assistant.
Data Usage & Privacy
Quantum Workplace uses data collected within the system to improve the ability to process text and inform the AI tools' outputs. However, Quantum Workplace does not use any third-party training, validation, or testing. All of these processes are done internally.
Quantum Workplace tracks the performance of the AI tools across the platform which requires an evaluation of the outputs provided to users. These evaluations inform product enhancements. Quantum Workplace's AI tools do not store any of the information they consume to ensure privacy and confidentiality.
Model Training
Data privacy and security are prioritized in the training of the Quantum Workplace AI models.
Models are not trained with sensitive or specific information, e.g. PII data.
The training of Quantum Workplace models happens on context-retrieval components within the environment. These context-retrieval components are used in AI to determine what information needs to be gathered to complete the request, given the available information.
AI Risks & Governance
Quantum Workplace has taken steps to address AI risks regarding accuracy/reliability, bias/discriminatory impacts, privacy, confidentiality, and transparency.
Quantum Workplace frequently validates the AI tools on a regular-basis and has a dedicated multi-disciplinary cross-functional team that oversees AI tools and risks.