Entrepreneurship Meets Artificial Intelligence at NJIT
The world of commerce was thoroughly represented at New Jersey Institute of Technology’s inaugural Artificial Intelligence Exploration Day, where several faculty and students presented their AI-enabled research covering topics from entrepreneurship to human-machine collaboration to real estate titling.
Presenters represented Martin Tuchman School of Management, the university’s traditionally tech-focused business school that evolved from coursework and student groups at NJIT predecessor Newark College of Engineering as early as the 1920s.
Innovation and business-building are favorite topics among MTSM faculty and students, and it showed through at AI Day. Associate Prof. Cesar Bandera, the Leir Endowed Chair of Entrepreneurship, presented on using AI to evaluate teaching performance of student entrepreneurial mindsets. Associate Prof. Raja Roy compared the current AI frenzy with the robotics trend of the 1980s, to derive lessons for entrepreneurial startups and diversifying firms.
Bandera teaches students to consider their own thoughts on entrepreneurship by drawing mind maps, which are informal thought-bubbles with branches and offshoots. He evaluates their maps based on five criteria — internal locus of control, intrinsic motivation, need for growth, resilience and self-efficacy — but he wondered if AI could perform such evaluation. Ideally, it would do so with equal accuracy and greater speed, he said in his presentation, Consistent or Capricious?.
“GenAI shows reliability for basic quantitative and qualitative analysis tasks and is potentially useful for large-scale initial assessments. However, caution and human oversight are necessary for sophisticated (e.g., model-based) assessment tasks, and AI-assisted assessment tools will require regular revalidation for longitudinal consistency,” Bandera wrote in a research paper on the topic. “The findings also have implications for the development of GenAI platforms. Developers conduct benchmark and safety testing on new generations of their platforms prior to release [however] longitudinal considerations may have been overlooked … testing should also evaluate temporal consistency in GenAI output.”
Roy, in Navigating the AI Gold Rushes, showed parallels between companies in the 1980s that made noise in robotics and those creating a stir in AI-enabled robotics today. He presented on robotics specialist Unimation and industrial giant Westinghouse, as compared to modern companies like BYD, Neralink, Optimus and Tesla.
Robots would replace skilled labor by the 1990s, prognosticators stated in the era of Reagan and home computers, although history shows a more cautious rollout. Now, for questions such as which traits distinguish successful companies, which capabilities truly matter for survival in AI and will there be a geographic shift from America to Asia, “Raising open-ended questions such as these is the purpose of the session. Having concrete answers to these questions is not,” Roy said.
In another presentation, PaperBasket: Saving the Research World (from Itself), Ron Bekkerman discussed how AI can assist peer reviewers for journal articles. Bekkerman, as director of MTSM’s Leir Research Institute for Business, Technology and Society, is often solicited to be a reviewer. But he said the deluge of AI-written papers and low-quality human papers has made the task nearly impossible. He worked with computer science students to build PaperBasket as their capstone project. It uses AI to search a four-million-strong online paper repository and tells reviewers if the topics and conclusions are novel or redundant. The same software could also be used by researchers themselves before embarking on new projects, or by journalists wanting to know if a topic is truly newsworthy.
Mark Annett, professor of practice, led his audience on a look at the Portable Human – AI Collaboration and the Future of Work. Regarding policies and technical matters, “In the future, human beings may not be hired as individuals alone, but as centaurs — human/AI pairs — because they are far more productive together than either alone. As AI becomes tuned to how you think and work, a critical question emerges: who owns that collaboration, and can you take it with you,” Annett stated.
Annett explained that portability would be necessary when someone switches from using ChatGPT to Gemini, or when they work for a different company or transfer to another college. His recent patent filing, Centaurial Processing Unit, allows for portability by extracting a user’s data from an AI system and auditing that information to remove anything proprietary.
“Collaboration behavior itself may be made portable across systems without exposing proprietary information. The goal is to help audiences see that centaur careers are not just a metaphor for the future of work, but a design problem that universities, employers, and technologists will increasingly need to address,” Annett noted.
A pair of students also made noteworthy contributions to AI in the business world. Frank Cordoba showed off Fortis Title Network. His software uses AI to modernize real estate title research and lien analysis. “Using natural language processing and semantic search, the project analyzes fragmented public records to improve accuracy, reduce risk and increase transparency in real estate transactions,” he stated. Meanwhile, Sarisha Singh presented on AI-Powered Risk Intelligence from Consumer Complaints in Fintech. In her project, “by combining topic modeling, semantic embeddings and controlled AI labeling, the project transforms unstructured text into explainable insights on consumer pain points and systemic risk.”