Empowering Computing: Stevens Researcher Pushes for a More Open, User-Centric Digital Future

For decades, computing has been shaped by corporate interests, restricting access to technology and limiting users' ability to control their own digital environments. But at Stevens Institute of Technology, Assistant Professor of Computer Science Michael Greenberg, an expert in programming languages and software systems, is working to reclaim computing for the people—one command line at a time.

Greenberg’s research in Stevens’ Department of Computer Science centers on democratizing coding, making computing more accessible and allowing individuals to navigate their digital environments independently. One recent project, MultiPL-E, is a groundbreaking initiative with researchers at Northeastern, Oberlin and Wellesley aimed at improving open-source artificial intelligence (AI) models for code generation, ensuring that technological advancements remain available to everyone—not just corporations with private datasets.

"We’re very interested in just getting things right," Greenberg said. "And the more we can get computing right, the more computing can serve us."

The challenge of closed-source AI

Generative artificial intelligence has revolutionized coding, but many of the most powerful models remain closed-source, controlled by a select few companies. Greenberg’s research has revealed a troubling divide: proprietary models such as OpenAI’s Codex and similar models vastly outperform their open-source counterparts, raising concerns about transparency, access and fairness in technological development.

"The large language model (LLM) that performs the absolute best in our paper on MultiPL-E—OpenAI’s Codex—is closed-source. Nobody knows what its training set is," Greenberg explained. "That’s a bummer in a lot of ways."

By limiting access to their data and models, private AI firms create barriers for independent researchers, educators and smaller developers. The implications extend beyond academia: lack of transparency raises legal and ethical concerns, particularly regarding copyright, data privacy and bias in AI-generated code.

The open-source alternative: a fairer future for computing

To counter this, Greenberg and his collaborators have helped develop BigCode, an ambitious open-source initiative designed to create freely available, high-quality AI models for code generation. Their contributions to BigCode have helped make it easy to generate code in less popular languages. By prioritizing ethical data sourcing and full transparency, BigCode ensures that academic and independent researchers have the same opportunities to innovate as corporate entities.

"All of this makes it so that it’s more of a level playing field for everyone working on large language models and generative AI," Greenberg says. “It’s nice to have [BigCode] be a bunch of people who are trying to do the right thing."

Reclaiming user control in computing

Beyond AI models, Greenberg’s passion lies in restoring users’ ability to manage their own computing environments. He draws inspiration from the early days of personal computing in the 1980s, when the goal was to empower individuals rather than make them dependent on software companies.

Assistant Professor Michael Greenberg

"There was this vision of computing where it was all about personal power," Greenberg said. "The idea was that you could manage your life, your business, your personal interests using computing. But now, software is designed to extract value from users rather than help them manage their lives."

One key to regaining this control, he argues, is mastering the Unix shell, the powerful command-line interface that serves as the backbone of modern operating systems. Understanding the shell means being able to customize and control computing experiences without being confined to prepackaged software solutions.

"Mastery of the shell is what lets people take ownership," Greenberg said. "If the shell is the control hatch, then you’re inside the machine—you’re in charge."

With a faculty dedicated to both cutting-edge innovation and the democratization of technology, Stevens and researchers like Greenberg are ensuring that the future of computing remains open, fair and accessible.

As AI continues to evolve, the question remains: who will control the tools that shape our digital world? Thanks to Greenberg and others, the answer might not be locked away behind a corporate paywall—it might belong to all of us.