AI meets the grade: Faculty reflections on Essaybot in the classroom
At the D'Amore-McKim School of Business, artificial intelligence isn't just a research topic—it's becoming part of the teaching toolkit. Powered by the DASH_box, Essaybot is an AI-assisted grading tool designed to help instructors provide faster, more consistent, and more thoughtful feedback on student writing.
This academic year, three D'Amore-McKim faculty members—Martin Dias, Sean M. Gallagher, and Zeynep Aksehirli—piloted Essaybot in their courses. Their experiences showcase both the promise and the practical considerations of integrating AI into grading and assessment, as part of the D'Amore-McKim AI Strategic Hub's broader mission to support innovation in teaching, research, business, and social impact.
Martin Dias
Balancing AI speed with human intuition
In a recent pilot study, Dias tested the AI grader alongside human assessment to evaluate its accuracy, efficiency, and overall effectiveness in providing student feedback. One key aspect of this study was the involvement of Kaitlyn Lai, a human TA/grader whose reflections offered critical insights into how AI and human judgment compare in essay evaluation.
One of the standout advantages of Essaybot is speed. Dias noted that once set up properly, the AI tool reduced grading time by 60%, potentially saving up to 80% in a fully optimized system. While the initial setup took a few hours, future uses could be streamlined to just one hour per assignment. “The grade review took about 10 seconds per instance, and feedback review about two minutes per instance. In total, Essaybot could save me 8 hours per exam,” he says. However, Dias acknowledged that fine-tuning is still needed to fully align Essaybot's grading with instructor preferences.
Another significant advantage of AI grading is consistency. Unlike human graders, who naturally experience fatigue, bias, or shifting judgment over time, Essaybot provides uniform evaluations across all submissions. Lai, the human grader involved in the study, says there's an emotional aspect to grading. “Even with students anonymized, I still felt ‘guilty' because I felt like I was being harsh. I would often compare the grade of the student I was currently grading to previous students' grades.” She also highlighted the risk of answer fatigue, where repeated grading of similar responses leads to unintentional leniency or harsher scoring due to cognitive overload.
The AI-human synergy in education
While Essaybot significantly improves efficiency, Dias noted the importance of preserving personalized feedback. Students may resist fully AI-generated feedback, requiring instructors to adapt AI responses to match their tone and teaching style.
Dias also raised a crucial long-term question: “Will AI automation reduce my empathy for students over time? If grading becomes purely mechanical, will I spend less time deeply considering student struggles? Perhaps this will free me to engage with students in more meaningful ways, such as mentoring and preparing them for co-op interviews. He envisions a future where AI grading incorporates tailored feedback based on student attributes, such as their major, co-op experience, and learning preferences.
Moreover, Essaybot has revealed how much tacit knowledge exists in teaching. Since AI relies on explicit, codified knowledge, future advancements may require a deeper integration of course materials and instructor intuition to enhance AI grading accuracy.
Sean M. Gallagher
Revolutionizing grading with AI
Gallagher says that grading essays is often one of the most time-consuming and challenging aspects of teaching. He highlights a common struggle: “If you're grading 30 essays, it is virtually impossible to provide consistently high-quality feedback from the first essay you review versus the 30th. Instructors do their best, but we're human, we tire, and we can get distracted.”
With Essaybot, this challenge is significantly reduced. Gallagher's pilot test demonstrated promising results, suggesting that DASH is on the right track with this innovative tool. He estimates that a fully developed version could reduce grading time by over 70%, a game-changer for instructors juggling multiple courses and large class sizes.
Could AI bring back essay-based exams?
Due to the labor-intensive nature of grading, many instructors, including Gallagher, have emphasized assessments that are traditionally more scalable with relatively less writing. However, Essaybot could change that trend. “Would the availability of a valid and reliable Essaybot tool change how I teach and grade? There is no doubt about it. I would utilize more essay-based quizzes,” he says.
If Essaybot continues to thrive, it has the potential to revitalize critical thinking assessments and improve how students engage with course material.
Beyond just grading, Essaybot represents a broader shift toward AI-assisted education, a movement that could help reduce costs and improve accessibility. “To lower the cost of future higher education, we must find ways to significantly boost institutional productivity while maintaining or enhancing quality,” Gallagher says. “A tool like Essaybot has the potential to increase instructor productivity substantially and help make higher education more affordable.”
Zeynep Aksehirli
Embracing AI to enhance teaching methods
With a background in engineering, Aksehrili's specialization is at the intersection of technological and human elements in an organization. She has been using EssayBot to provide feedback on student writing in her writing-intensive business analysis course. Her experience highlights several key benefits of integrating AI into the grading process:
- Faster feedback turnaround: “The way I use the bots is increasing the efficiency of the feedback I'm giving, so we can turn things around quicker. This means we can do more iterations in terms of learning and (teaching) different ways of writing,” she says.
- More learning iterations: The faster processing allows for multiple rounds of feedback, providing students with more opportunities to improve their writing skills throughout the course.
- More personalized feedback: The efficiency of AI enables her to provide more detailed and individually tailored feedback to each student.
- Better tone in feedback delivery: She notes that sometimes instructors give quick and abrupt feedback simply because they have limited time, but AI has the efficiency to provide much smoother feedback that would potentially be perceived better by the student, or as she put it, “more growth-oriented rather than critique-oriented.” However, she also notes that AI may sometimes hold students to a more rigorous standard.
- Consistency in grading: The experiment shows notable consistency between human and AI assessment levels, creating reliable student evaluation standards.
- Integration as a tool rather than a threat: Regarding AI's broader impact on education, Aksehrili appreciates the supportive approach of the DASH initiative. “I was happy when the discussion was about ‘how do we integrate everything into both our work and our students' work in terms of a tool,' rather than not (being) scared of it.”
- Time savings for higher-level tasks: She views AI as a powerful efficiency tool that allows educators to focus on more creative work. “AI is just an excellent way of fitting data into a curve,” she says. “So whatever it's producing is content based on what it knows. And our skill as humans is supposed to be going beyond that, going further.” This philosophy guides her use of AI to handle routine tasks, allowing her to dedicate more time to creative and innovative teaching approaches.
Looking ahead, Aksehrili envisions AI becoming fully integrated into academic life: “The DASH bots are going to help make the more routine parts of our lives easier, quicker and more to the point, so that we can use the time saved, if you will, in creating new content, creating different kinds of ideas.”