Jump to page content

Hard AI Institute

Purdue's new Institute of Hard AI connects AI with the physical world.

Banner image: Purdue’s College of Engineering has launched the Institute of Hard AI, establishing Purdue as a national leader in artificial intelligence at the interface of the virtual and physical worlds.

By Brian Huchel

Purdue’s College of Engineering announced the formation of the Institute of Hard AI in June 2022 during the 11th annual meeting of the National Academy of Inventors in Phoenix. The college-wide institute positions Purdue as a national leader in artificial intelligence at the interface of the virtual and physical worlds.

To understand the term “hard AI,” think of it as bytes meet atoms. The full power of AI is at the interface between what we code and what we touch.

Intended to become a national leader at this frontier and guided by a top-tier alumni advisory board, the institute will combine research, teaching and commercialization in three areas:

AI BY HARDWARE: Future AI applications will place unprecedented demands on computing platforms, requiring advances in function-specific AI hardware to control complex systems and interconnected processes. Autonomous and connected systems, energy generation/distribution, healthcare and smart agriculture are examples of systems that will require specially designed AI chips. Leveraging Purdue’s unique strength in cognitive computing, the Institute of Hard AI will focus on a unique algorithm-to-hardware codesign approach, addressing challenges at all levels: algorithms, system architecture, circuits and devices.

AI OF HARD STUFF: The location for AI is increasingly dispersed through edge computing, where computation, communication and control all move closer to the end users and are dispersed in a distributed, variable, highly constrained and underorganized computing substrate. This brings new challenges to the design of AI systems, including hardware heterogeneity, robustness and the need for extreme energy efficiency. Purdue has some of the world’s leading experts in edge computing, remote sensing and 5G systems. AI doesn’t just live in the cloud — it will be all around us.

AI FOR PHYSICAL THINGS: When AI is used for decisions at the interface of virtual-physical worlds, the criteria for success change. For example, responsiveness and safety matter no less than accuracy and throughput. From autonomous trucks to pharmaceutical manufacturing, from nuclear reactor operation to connected things in digital agriculture, Purdue leads the nation anytime applications touch physical things.

Kaushik Roy, the Edward G. Tiedemann Jr. Distinguished Professor of Electrical and Computer Engineering, will serve as director of the institute. An alumni advisory board is comprised of luminaries from industry, including Chris Lister (BSIE ’95), Mike Harris (BSCEE ’91), Dan Rosckes (BSIE ’84) and Ujjwal Singh (BS Computer Science ’94 and BSEE ’94).

The institute’s initiatives include cutting-edge research and innovation; courses related to hard AI for both industry and academia, including a minor degree in AI for physical applications; a distinguished lecture series; industry partnerships; student-industry networking and career development; and technology commercialization. The effort will also grow to include multiple other departments and colleges on the Purdue campus.