Engineering a future for reliable AI

Data science doctoral student Mohammad Sadeq Abolhasani builds profile as a researcher who can turn information into actionable knowledge

Information is everywhere. But how can huge amounts of data be used to help people acquire actionable knowledge?

This is the central question driving the research of Mohammad Sadeq Abolhasani. He is a data science doctoral student in the School of Computing and Augmented Intelligence, part of the Ira A. Fulton Schools of Engineering at Arizona State University. Working under the supervision of Rong Pan, a Fulton Schools professor of industrial engineering, he studies artificial intelligence, or AI, especially the of use large language models, or LLMs, where computers are trained to understand human language by scanning huge amounts of text. Abolhasani’s work focuses on how to boost the reliability of such systems.

As many industries increasingly incorporate AI into their operations, an urgent need to ensure that users can rely on this new technology is developing. This emerging area of research is of interest to the expert organizers of the Annual Reliability and Maintainability Symposium, or RAMS.

The annual symposium is hosted by seven technical societies, including the Institute of Industrial and Systems Engineers and the IEEE Reliability Society. In January, Abolhasani received the prestigious Hans Reiche RAMS 2025 Scholarship and was invited by symposium organizers to present his work at the event.

A graphic illustrating how Abolhasani’s algorithms generate knowledge graphs. A knowledge graph provides a computer system with a visual representation of the data to which it has access, helping the system understand the relationship between different types of information. Here, a system has first been trained to understand the general concept of book publishing (shown left). Then, it applies what it understands to data about the career of novelist Harper Lee (shown right). The result will be a tool that answers specific questions about the timeline, locations and publishers of Lee’s books. Graphic courtesy of Mohammad Sadeq Abolhasani

Knowledge is power

Imagine a trio of new engineers hired by a manufacturer of technical equipment to make improvements to its products. They need to be quickly brought up to speed. The company has an enormous amount of information stored in documents, videos and images, but the employees need to get to work immediately. The manufacturer wants an AI system to comb through the data and answer the engineers’ questions. But the answers must be correct.

“Right now, many people are tapping into large language models with tools like ChatGPT. These contain a lot of general information,” Abolhasani says. “But industrial users need to be able to mix this general information with domain-specific information relevant to their work.”

Abolhasani’s research seeks to address those very issues. He develops AI-powered algorithms, the sets of instructions that computers use to do their work, that create knowledge graphs. These are visual representations of data that clearly show the relationship between different pieces of information. This is especially helpful because data is increasingly stored in a variety of forms, including images and video.

In the hypothetical scenario above, the manufacturer’s AI system might scan internal technical videos, industry reports about market trends and insights from social media images, combining general and specific information that could help the engineers make informed decisions about product design. The process could allow experts to quickly turn information into real knowledge.

The overarching concept is to produce AI systems that can reason like humans do and produce insights that users can trust.

At RAMS, Abolhasani gave a 30-minute invited talk, served as a moderator at a panel session and presented his work at a poster session. He also outlined next steps in the project, which includes the use of advanced forms of AI agents to make the learning process even more efficient for users.

A data-driven future

Abolhasani previously earned his undergraduate degree in industrial engineering from Qazvin University and his master’s degree in industrial management from the University of Tehran. He is a past recipient of the Fulton Fellowship Award.

During his doctoral studies, he also seeks to cultivate key industry relationships, serving in various data science roles for companies, including IFS, a global provider of software platforms that helps businesses manage their operations and resources.

Pan, who played a key role in the development of the data science doctoral degree offered by the School of Computing and Augmented Intelligence, says Abolhasani’s work exemplifies what the program hopes to accomplish.

“We’re striving to prepare students for high-level and leadership roles across industry and academia,” Pan says. “Mohammad’s recognition is a testament both to his own level of scholarship as well as the overall quality of data science work being done here at ASU.”

Abolhasani hopes to leverage his research accomplishments into a career making impactful contributions to data science.

“For me the goal is to create something really powerful and meaningful, a real product for industry,” he says. “Data science is a tool for everything. There is almost nothing that cannot benefit from data science these days.”