Emre Kiciman: Advancing AI for Industry and Society at Microsoft Research

Emre Kiciman: Driving AI Innovation at Microsoft Research

This document provides a comprehensive overview of the work and research interests of Emre Kiciman, a Senior Principal Research Manager at Microsoft Research Redmond. Kiciman leads the AI for Industry team within Research for Industries, focusing on the large-scale application of AI, its societal impact, and the development of causal machine learning algorithms and generative AI systems.

Research Focus Areas

Kiciman's research spans several key areas within artificial intelligence and machine learning:

Causal Machine Learning: A significant portion of his work involves broadening the use of causal methods for decision-making across various application domains. This includes scaling up conventional causal inference techniques to handle large and high-dimensional datasets, adapting these methods to new settings, and improving the robustness and bias of prediction and classification algorithms through causal or causal-inspired approaches.
AI's Societal Impact: Kiciman is dedicated to promoting the positive applications of AI while mitigating its negative implications. His projects address the intersection of security and machine learning, exploring new attacks and defenses for security-critical AI systems. He also investigates data biases, privacy-preserving AI systems, misinformation, and the ethical considerations surrounding AI.
Computational Social Science: With a strong interest in computational social science, Kiciman focuses on social media analyses that require a causal understanding of phenomena in health, mental health, data bias, and how new technologies influence our perception of the world and information discovery.
Distributed Systems and Information Retrieval: His past research includes work on the reliability, architecture, and operations of distributed systems, applying machine learning to fault detection and diagnosis. He has also contributed to monitoring and optimizing web applications and various information retrieval tasks, such as entity linking and using social context for document ranking.

Key Projects and Highlights

Several notable projects and publications highlight Kiciman's contributions:

AI Controller Interface: This project introduces a framework that simplifies the creation and experimentation of new strategies for improving Large Language Model (LLM) generations through AI Controllers. These controllers integrate with the LLM inference engine, enabling constrained decoding, dynamic editing of prompts and text, and coordination across multiple generations.
Causal Reasoning and Large Language Models: This research explores the causal capabilities of LLMs, examining their implications for critical domains like medicine, science, law, and policy. It delves into different types of causal reasoning tasks and the challenges of construct and measurement validity.
DoWhy and PyWhy: Kiciman has been instrumental in the development of DoWhy, a Python library for causal inference, which has evolved into the independent PyWhy model. This initiative aims to foster the growth of causal inference as a field.
Foundations of Causal Inference: This work focuses on the fundamental principles of causal inference and its impact on machine learning, emphasizing its importance for decision-making and improving outcomes.
Social Data: Biases, Methodological Pitfalls, and Ethical Boundaries: This research addresses the complexities of using social data in digital applications, highlighting potential biases, methodological challenges, and ethical considerations.

Research Areas and Groups

Kiciman's work aligns with several key research areas at Microsoft Research, including:

Artificial Intelligence
Human-Computer Interaction
Search and Information Retrieval
Social Sciences
Systems and Networking

He is also associated with the following research groups:

Causality and Machine Learning
Research for Industry

Contact Information and Social Media

Email: emrek@microsoft.com
Websites: http://x.com/emrek, http://kiciman.org/
Google Scholar: https://scholar.google.com/citations?user=QZCU3NkAAAAJ&hl=en&oi=ao

Kiciman is active on social media platforms, sharing his work and engaging with the research community:

X (formerly Twitter): @MSFTResearch
Facebook: microsoftresearch
LinkedIn: Microsoft Research
YouTube: Microsoft Research
Instagram: msft_research
RSS Feed: Available for updates.

Sharing and Privacy

The page also includes links for sharing the content on social media platforms like X, Facebook, LinkedIn, and Reddit. It also provides information on consumer privacy choices and Microsoft's privacy policies.

Related Microsoft Products and Initiatives

The content touches upon various Microsoft products and initiatives, including:

Surface Devices: Surface Pro, Surface Laptop, Surface Laptop Studio 2, Surface Laptop Go 3.
AI and Cloud: Microsoft Copilot, AI in Windows, Microsoft Cloud, Azure, Dynamics 365, Microsoft 365, Microsoft Advertising.
Developer Tools: Developer Microsoft, Microsoft Learn, Azure Marketplace, AppSource, Power Platform, Visual Studio.
Other: Windows 11 Apps, Microsoft Store, Education initiatives, Careers, About Microsoft, Sustainability.

Visuals

The page features several images and graphics related to Kiciman's research, including:

A banner image related to privacy choices.
A portrait of Emre Kiciman.
Featured project images for AI Controller Interface, Causal Reasoning and LLMs, DoWhy, Foundations of Causal Inference, DoWhy: Causal Reasoning, and Social Data.
Icons representing different research areas.