Peter Hess

Alexandre Choquette

Ashley Peterson

Mike Murphy

Kim Martineau

News

Research

Q & A

Release

Explainer

Meet AI-Hilbert, a new algorithm for transforming scientific discovery

Teaching AI models to improve themselves 

For LLMs, IBM’s NorthPole chip overcomes the tradeoff between speed and efficiency

Accelerating discoveries with new AI tools, powerful chips — and turkey

IBM and Université de Sherbrooke announce two quantum research chairs

IBM and ESA open-source TerraMind, the best performing generative AI model for Earth observation

Miklós Ajtai and Regina Barzilay received IEEE medals for applying unexpected methods to computer science

Eurocrypt 2025

ODSC East 2025

DAC 2025

ECTC 2025

Canadian Separations Symposium 2025

Trustworthy AI

Trustworthy Generation

Quantum Safe

Natural Language Processing

Generative AI

Exploratory Science

Physical Sciences

Computer Science

Quantum Community

Semiconductors

AI Hardware

Explainable AI

Technology impacts every aspect of our work and lives. It is imperative that technology is built responsibly to reduce differential effects on various populations. At IBM Research, we are working to ensure that the technologies we create promote beneficial outcomes for our clients and for the world.

Responsible Technology

IBM has been at the forefront of computer science research since the field’s inception. Our research today focuses on achieving breakthroughs in automation, information processing, and computation. Our goal is to complement and extend human performance — and advance society as a whole. 

We’re working on automation, information processing, and computation research to complement and extend human performance and advance society.

Our long history of research has had an enduring impact on computer science, operations research, and information theory. We’re currently focused on optimization, probability, complexity, geometry of data, as well as linear and multi-linear algebra, to deliver tools that are fundamental to big data and AI.

Mathematical Sciences

We’re focused on foundational mathematical research with the aim of delivering tools to that are fundamental to big data and AI.

Our physical sciences researchers seek to understand materials and their properties through advanced characterization methods, explore novel devices and materials for new computational platforms, and apply physics to improve AI algorithms and interpretability.



We’re using physics to improve AI algorithms and interpretability, as well as novel materials for new computational platforms. 


Our researchers seek to answer some of the most fundamental questions. We strive for a greater understanding of how our world works and how we can harness that knowledge to advance science, mathematics, and computation.

Mathematical constants and variables in white boxes float on a black background, connected with green arrows

IBM Research scientists are working on different approaches to teach AI models how to improve themselves

A 2U server blade containing NorthPole cards

Images of the IEEE Frances E. Allen Medal (left) and the John von Neumann Medal (right)

Exploratory Science

Our work

Miklós Ajtai and Regina Barzilay received IEEE medals for applying unexpected methods to computer science

IBM and ESA open-source TerraMind, the best performing generative AI model for Earth observation

IBM and Université de Sherbrooke announce two quantum research chairs

Accelerating discoveries with new AI tools, powerful chips — and turkey

For LLMs, IBM’s NorthPole chip overcomes the tradeoff between speed and efficiency

Teaching AI models to improve themselves

Topics

Publications

MDLab: AI frameworks for Carbon Capture and Battery Materials

CIRCUITSYNTH-RL: LLM-Based Circuit Topology Synthesis with RL Refinement

Electrical Performance of Hybrid Bonding with Sub-Micron Cu-Cu Bonding Contacts: Effects of Scaling, Microstructure, and Surface Morphology

Guardrails in generative AI workflows via orchestration

How AI can accelerate Materials Discovery for CO2 Capture

POKE: A Compact and Efficient PKE from Higher-dimensional Isogenies