Last year, mechanical engineering professor Andrew Gadsden worked with NASA to send a self-retracting robotic mount 95 percent above Earth’s atmosphere.
So it’s no surprise that he’s currently working on a new partnership with a Toronto-based data science and artificial intelligence (AI) consulting firm called Adastra, which means “to the stars” in Latin.
Funded through a four-year joint grant supported by the NSERC Alliance and Mitacs Accelerate, Gadsden’s project explores two rapidly evolving areas of AI: explainable AI and agentic AI, with the goal of creating intelligent systems that are not only capable, but also transparent and trustworthy.
No more black boxes
One phase of the project involves working with agent AI to deploy reliable systems across a variety of industries including healthcare, finance, manufacturing, and even space.
Agentic AI is an autonomous system that operates under limited human supervision and is responsible for making decisions and adapting to changing environments.
An example of this type of system is automated fraud detection, Gadsden explains. Deploying AI agents can flag suspicious charges as fraud, gather and gather relevant supporting information about fraud, compare similar claims, and automate the generation of audit reports.
But with limited knowledge of how the system works and minimal human interaction, how can users trust that an agent has correctly identified a claim as fraud?
This is where the second phase of the project, Explainable AI (XAI), comes into play.
XAI aims to build trust and trust in AI systems among users. Current AI models include what are called black boxes, allowing users to see what goes into the system (inputs) and what comes out (outputs), but not the inner workings that make the system work.
Gadsden explains that the ability to see inside the black box allows users to better understand the overall system, have more confidence in what the AI model is telling them, and regularly monitor the health of the system. Additionally, XAI shows the functional boundaries of the underlying models, tools, and agent processes, increasing transparency and supporting building more robust and accurate solutions.
“The future of AI is about more than just smarter machines,” Gadsden said. “It’s about building systems that we understand, trust, and can responsibly integrate into the world around us.”
An ecosystem growing right here at home
For Gadsden, this work with Adastra is more than just a single research project, it reflects a larger picture that is taking shape across campus.
At McMaster, he explains, he helps shape the way researchers develop and deploy AI beyond the university, ensuring innovations are aligned with critical thinking and real-world impact.
“Leading in innovation is more than a single-threaded effort; it requires real-world applications, corporate discipline, and intentional experimentation,” said John Yawney, chief analytics officer and industry partner at Adastra.
“Our partnership with McMaster allows Adastra to responsibly scale innovation while increasing both our professional and technological capabilities to meet tomorrow’s challenges.”
And with research programs across engineering, health sciences, business, and other disciplines, Gadsden explains, McMaster is rapidly becoming a national hub for applied, interdisciplinary AI research.
“What’s happening at McMaster is bigger than just individual research programs,” Gadsden says. It’s also important to use AI as a tool to accelerate discovery itself, rather than just inventing new AI models and frameworks.
For Gadsden, the amount of AI work happening on campus shows a great desire to explore AI’s potential in discovery.
“The work we are doing in this area is changing the landscape for researchers and students alike,” said Carlos Filipe, associate dean for research, innovation and partnerships in the College of Engineering.
“This creates an opportunity to collaborate across disciplines, tackle real-world challenges, work closely with industry partners, and shape the future of technology.”
Open the door to the next generation
Beyond the applied implications of this research, there are also very real risks for current and incoming graduate students.
As AI is reshaping the world’s workforce, we are working directly with data analytics companies to build intelligent systems and define how they are trusted, putting McMaster graduate students at the forefront of the transformation field.
And Gadsden, who also serves as vice chair for graduate studies in mechanical engineering, is paying attention.
“This project is at the intersection of mechanical engineering, computer science, and AI,” Gadsden explains. “This allows our students to make the most of all these worlds.”
Those who participate in the project gain more than technical expertise. You will gain experience collaborating with industry partners to communicate complex ideas to non-academic audiences and see your research have real-world impact.
“They leave with advanced degrees, specialized training, and strong professional networks,” Gadsden points out. “This combination opens doors for industry and academia.”
This was also true for Naseem Al-Saadi, a former PhD student in Andrew Gadsden’s lab who collaborated with Adastra during his studies and now works for the company.
“In addition to being research-intensive, I also gained valuable industry experience at Adastra, which accelerated my graduate studies at McMaster,” Al-Sadi says.
“This unique experience really helped me get a head start on my career.”
Students don’t have to come in as experts in every field, Gadsden said. “The most important things are curiosity, motivation, and a desire to learn.”
