In this work in progress we explain the development of a process or changing the teaching culture at the department of mechanical engineering of Texas A&M. The Strategy is based on addressing (a) development of reflective teachers (b) developing shared vision (c) considering policy and (d) faculty initiated curricular innovations. Teachers were trained in lean startup (incremental innovation) approaches to educational innovation. A community of practice was developed to build trust, share learning from innovations and to develop accountability. A systematic approach is being developed to evaluate the efficacy of this change strategy especially with engaging late adopters and sceptics. Preliminary results indicate that that creating and nurturing a teaching community of practice (a part of a comprehensive change strategy) is very effective in promoting a culture of innovation in teaching.

We investigated the capability of Large Language Models (LLMs) for grading short answer questions and studied different auto-grading schemes for evaluating student responses to conceptual questions in a mechanical engineering statics course. We compared the ability of seven Natural Language Processing (NLP) systems to score text-based answers as Correct/Incorrect and numerically, with human-supported Rules-based grading as a benchmark. We collected the instructor-provided answers, anonymized student answers, and their grades for this study. The findings reveal that the Large Language Model (LLM) based grading systems exhibit commendable precision in binary evaluations. However, within the spectrum of error classifications, the LLM-based grading systems exhibit a pronounced rate of false positives, a scenario less than ideal in an educational context. Considering that the technical terms in the instructors' answers are a primary factor in grading, our forthcoming research endeavors to embed keyword detection within the LLM-based automatic grading framework to mitigate the incidence of false positives. Thus, we investigated the ability of the standalone LLM-Vicuna to identify important keywords in an answer in the context of the mechanic's course. Our preliminary observations indicate that Vicuna accurately identifies the keywords in the answers, but the results are not yet repeatable due to the stochastic nature of the model.

How innovative are you? High levels of innovativeness are correlated with high-value creation, and today's rapid pace of change requires new skills for lifelong learning, curiosity, innovation, and continuous improvement. To what level can personal innovativeness be improved through curricular innovations and how do we improve students' practical skills for ongoing discovery? Innovation Mind and Skill Sets for Design and Research is a new course that focuses on providing STEM students the broader mind and skill sets that will help them connect their specialized work to the larger system and the society it impacts and effectively experiment, discover, and create value in the highly dynamic and uncertain environment of breakthrough innovation. This paper will discuss the development of the course which Hipwell based on both innovation literature and her own two decades of experience applying and improving innovation business processes with her teams in a fast-paced, high-tech industry. An educational class structure and learning outcomes are developed based upon the theory that innovation is itself a learning process and can be improved through the known learning science concepts of improvement of student metacognition about their innovation process and incorporation of intention and reflection in their learning cycles. Further, the model incorporates literature on teaching system thinking and Transformative Learning Theory for adults. Content is developed that teaches the students practical techniques that scaffold their innovation process and increase their metacognition. Movement up levels of Bloom's Taxonomy and long-term habits are developed through the practice of the innovation process in a team project with expert feedback from an experienced innovation practitioner. Learning is increased through reflection on the process during class presentations. Measurement of student improvement results compared to control design curricula using the Innovator Mindset® Assessment is also discussed.