As business school faculty look to utilise technology for the benefit of students, ESSEC Business School Professor Fabrice Cavarretta has developed a new tool at the intersection of AI and pedagogy: an avatar of himself that can answer student questions.
We spoke with Cavarretta – a professor in Management and Entrepreneurship who teaches mainly the Leadership and Entrepreneurial Manager course in master programs, coordinates the PhD Entrepreneurship seminar and recently developed a Corporate Venturing course for executives – to find out more. Here’s what he had to say:
What was your inspiration for creating your “e-professor”? What’s the goal?
As an academic, I have always been interested in not only accumulating knowledge but also disseminating it. This has been a common practice throughout history, from the Middle Ages when books became popular tools for accumulating and disseminating knowledge to modern times when blogs and videos are used for the same purpose.
I was intrigued by the idea of using AI to play a similar role in knowledge dissemination. With my own knowledge spread across various documents, including practitioner articles and research papers, I wanted to see if AI could provide access to this corpus and answer questions based on it.
This led me to implement a system, which I call an “eProf Cavarretta”–to distinguish it from the physical Professor Cavarretta. I thereafter fed a “GPTs”, i.e., a customised version of ChatGPT, with my entire corpus and provided very specific instructions on how to answer.
The goal is to have the algorithm answer questions with a specific spin, similar to how I would explain concepts to my students. I believed that people would be interested in hearing the expert’s perspective rather than a general answer generated by a search engine or a generic algorithm.
There are many usage scenarios for a tool like this.
One scenario is when someone wants to know about my posture, specific teachings, research interests, etc., but they are not my students or even a student of my school. They can go to the eProf and ask the algorithm a question to get an idea of how I, Fabrice Cavaretta, the professor, would answer. They could find the eProf on my website or on the school’s website.
Another scenario is when I have a student in one of my courses that wants to clarify some things. Maybe they don’t want to disturb me during class, or there isn’t enough time because of the number of students. They can ask the eProf to see how I would answer, based on my unique perspective as their teacher.
A third typical scenario is when my former students, who have seen me in the classroom, remember how we dealt with things but now have new problems or want a refresher. They can use the GPT to ask questions to me as a professor in a convenient and cost-effective way.
These are scenarios I cannot practically deliver to the world, as I cannot answer all the questions that come up during my teaching or in the minds of my future and former students.
Having an algorithm that can smoothly deliver my knowledge, my postures, all these scenario is truly amazing.
AI tools are developing rapidly, what do you think are some of the benefits and risks that come with these tools?
AI, as we experience it today, is a remarkable technological revolution comparable in scale to the introduction of modern computing. The significance of AI lies in its ability to interpret information in a qualitative and nuanced manner, resembling human understanding. This has been a long-awaited development in the field of information technology. The potential of AI is undeniable, and it is expected to permeate various aspects of our lives.
While the positive aspects of AI are astounding, there are also significant concerns. One major concern is the potential for large organisations, whether governmental or not, to control these algorithms, thereby putting individuals at risk of oppression. To mitigate this risk, it is crucial that we regulate the use of AI in our societal interactions. We should consider allowing AI to perform many tasks but with limitations and controlled boundaries.
Another subtle concern is the increasing reliance on machines for our daily interactions. Many of our interactions, such as contacting firms, call centres, administrations, etc., are already mediated by machines. With the advancement of AI, there is a possibility of ever more present complex bureaucratic scenarios, reminiscent of Kafka’s world. However, it is important to note that Kafkaesque bureaucracy existed before the advent of AI.
How do you think tools like this one could affect the business education space?
The benefits of AI for business education or education in general are immense. The technology now allows for natural interactions and complex understanding by the algorithm of human thinking and knowledge.
We can envision a system where an algorithm can guide students or executives in acquiring knowledge and practice. This is highly feasible, illustrated by the features offered by OpenAI, such as GPTs I use to build eProf Cavarretta. It enables the creation of customised environments that can manage complex scenarios.
For example, as a professor, I can create e-instructors that target specific students and guides them according to specific logic and timeline. The algorithm will take care of this process. One key advantage of these algorithms is their infinite patience, which is a game changer in the long term. While humans cost money, algorithms are more cost-effective for those who pay for them.
Furthermore, AI allows for one-on-one interactions, eliminating the need for traditional classrooms and enabling scalability.
All of this has obviously the potential to revolutionise the way education is organised.
Your eProf is an impressive tool. Do you think ChatGPT could be used by other professionals to create similar learning tools?
The tool is undeniably impressive, but it does have its limitations. Currently, there are still quite a few bugs related to the amount and type of documents that can be uploaded. Additionally, there is uncertainty about whether the algorithm can be instructed to always cite the sources provided by the user, an important point to consider in a pedagogical interaction.
The tool is designed to become even more impressive; OpenAI obviously betting a lot on it. This is only the first generation of the tool. Considering this, the potential for future improvements is immense.
For other professionals, I would recommend conducting experiments like this. One of the main reasons is to explore what unique contributions humans can make … once machines possess such capabilities. Although the advent of LLMs has significantly moved that boundary, I still believe some human interactions cannot be substituted for at least a few years. Humans possess the ability to sense and react to certain nuances that current classes of algorithms will find challenging to manage and predict.
Therefore, I am constantly contemplating my role as a teacher in a future where I can delegate a significant portion of knowledge dissemination. What would be then be my in-class value-added? In my case, I focus my teaching of leadership and entrepreneurship on the students’ relationship with the subjects and how they approach them.
This is something that I am uniquely qualified to do in a classroom, as it requires detecting emotions, resistance, and doing social engineering to overcome them. While basic knowledge can easily be acquired through reading books, blog entries, or step-by-step algorithms for business plans, the students’ personal connection and motivation towards subjects will remain areas where human experts will remain competitive compared to algorithms for the foreseeable future.
You are a professor of management, leadership and entrepreneurship. Do you believe that leaders could benefit from using AI tools? How?
Business leaders will be greatly impacted by the availability of new classes of tools. While IT has traditionally been used for calculations, technology now allows for the delegation to machines of tasks that require mild intellectual tasks. This includes gathering and summarising information, as well as detecting patterns.
The integration of AI technologies in various processes, such as call centres and customer interactions, will lead to significant changes in the nature of these processes and their associated costs.
As AI algorithms continue to advance, business leaders must be prepared to incorporate them as a key component of their endeavours. In the past, humans were the only universal input that business leaders had to deal with. However, we are now entering a new age where algorithms and artificial intelligence are becoming equally universal.
It is essential for business leaders to be able to effectively navigate and utilise these technologies. Even more so, given that such use of AI in all businesses process will raises issues, not always have positive consequences…
For additional insights, trends and perspectives on the use of AI in business education, visit the conversation here.