AI for Domains (AID)
Our mission
The M365 Research organization’s AI for Domains (AID) team is an applied research group focused on leveraging cutting-edge research to extend the capabilities, efficiency, reliability, and reduce risks associated with Copilot. We prioritize user privacy and confidentiality while unlocking the potential of enterprise data for domain-specific AI applications.
Our research
Customization and Domain-Tailoring
Enhance AI models’ relevance and utility across diverse domains by tailoring them to understand and interpret domain-specific data and challenges. This involves developing methodologies to customize LLMs with persona-specific knowledge, enabling them to deliver insights tailored to the unique contexts and requirements of different domains.
Trustworthy Grounding
Emphasizes the importance of basing AI decisions on reliable, factual information sourced from verifiable and authoritative databases, industry standards, and scientific research. This research area aims to improve the trustworthiness and credibility of AI-generated content, reducing biases and inaccuracies.
Privacy
Addresses the challenges of adhering to complex and ever-changing global regulations concerning data privacy, security, and ethical AI use. This research area ensures that domain-adapted models are not only compliant but also designed with the foresight to adapt to future regulatory changes.
Projects
Privacy Preserving Machine Learning Innovation
Recent research has shown that deploying ML models can, in some cases, implicate privacy in unexpected ways. For example, pretrained public language models that are fine-tuned on private data can be misused to recover private information, and very large language models…
dp-transformers
Training transformer models with differential privacy Transformer models have recently taken the field of Natural Language Processing (NLP) by storm as large language models based on the transformer architecture have shown impressive performance across a wide range of applications. However,…
Confidential AI
Our goal is to make Azure the most trustworthy cloud platform for AI. The platform we envisage offers confidentiality and integrity against privileged attackers including attacks on the code, data and hardware supply chains, performance close to that offered by…
Latest news
Research Focus: Week of June 10, 2024
June 12, 2024
Rethinking trust in direct messages in the AI era
September 5, 2023