AI as a Catalyst for Innovation: An Interview with Dr. Elliot Bendoly
AI, Mental Models, Double Loop Learning, and Symbiotic Learning Mechanisms
Introduction
Among its promising potentialities, generative AI possesses the capability to structure and influence our thought processes, making metacognition tangible. Yet, it also harbors a substantial risk: that its overuse, particularly by younger learners, might dilute their critical thinking abilities and dampen their creative autonomy. Navigating between these tensions is our ongoing challenge in incorporating generative AI into contemporary educational environments.
In January, Michael Spencer from AI Supremacy featured a guest post of mine on LinkedIn where I enthusiastically advocated for the integration of AI technology in education, urging educators to embrace this period of disruption as an opportunity for experimentation and to normalize AI in contemporary classrooms. Responding to my post, Dr. Elliot Bendoly shared a link to his recent research paper, “The role of generative design and additive manufacturing capabilities in developing human–AI symbiosis: Evidence from multiple case studies,” published in Decision Sciences in October 2023, co-authored with several colleagues. His study provides empirical support to my hypothesis, suggesting that a thoughtful implementation of AI has the potential to significantly enhance the educational landscape.
Since last summer, I've advocated for a data-driven approach towards AI's adoption and integration in education. At that time, our resources were scarce, primarily drawing from collective anecdotes and a vision gradually taking shape through online forums, training sessions, and conferences. However, I'm thrilled to share that the scenario is changing, with research papers beginning to emerge.
This development is incredibly exciting for us as educators, offering us solid evidence to guide our next steps. Many of us pioneering AI in education often work in isolation, but now we have access to comprehensive, meticulously conducted research like that of Dr. Bendoly's. This empowers us to develop and refine best practices for our professional routines and for the benefit of our students.
Research Summary
Dr. Elliot Bendoly's research explores the response of 48 companies to the COVID-19 pandemic, specifically focusing on those utilizing additive manufacturing (AM), or 3-D printing technology. Half of these companies adapted by embracing (AM) applications during the pandemic. Employing Additive Manufacturing (AM) entails engaging with intricate computing systems equipped with sophisticated AI processes.
This strategic pivot to AM and AI generally led to enhanced innovation, improved manufacturing efficiency, and better navigation through pandemic-induced challenges. Dr. Bendoly describes GD as "a symbiotic learning mechanism" and illustrates its impact through the concept of a "double learning loop." This concept highlights a two-step decision-making process.
First, designers select from the options that GD generates. Then, they must rationalize their choices within the context of these options. This reflective process deepens their understanding, enhancing their decision-making skills and expanding their design thinking. Dr. Bendoly highlights two major benefits of this approach: it shortens the time between decision-making and implementation, and it "has the potential to expand mental models in design, what works, and why."
Source: Online.wiley.com
Author Introduction
Source: Osu.edu
Dr. Elliot Bendoly stands out as a distinguished professor in Operations and Business Analytics at the Fisher College of Business, recognized for his significant contributions to the field, notably as the 2015 OM Distinguished Scholar by the Academy of Management.
His academic foundation is robust, with a PhD and MS in operations and decision technologies from Indiana University, complemented by degrees in economics and materials engineering from Case Western Reserve University. Before joining Fisher, Dr. Bendoly made his mark at Emory University’s Goizueta Business School and contributed as a visiting researcher at IE Business School in Madrid.
Dr. Bendoly's research intricately weaves together operations, information technology, and psychology, focusing on collaboration, complexity, and strategic alignment. His prolific output includes numerous articles in top-tier journals and influential books such as Excel Basics to Blackbelt and Visual Analytics for Management, which serve as key resources in management education.
An acclaimed educator, Dr. Bendoly has earned accolades for his teaching, including the 2014 Crystal Apple Teaching Award from Emory University. His commitment to education extends beyond the classroom, with valuable tools and resources available at www.blackbelt-apps.com.
Dr. Bendoly's editorial contributions to leading journals underscore his role in shaping academic discourse in operations management. His work bridges theoretical insights with practical applications, solidifying his reputation as a leading figure in the intersection of operations, technology, and organizational behavior.
Questions
1. Dr. Bendoly, in your research, you've adopted the "grounded theory-building method." Could you explain to my readers what this methodology entails and highlight how it uniquely positions your study?
Grounded theory-building refers to the identification of general patterns in things we observe or data we have access to, and the development of theoretical explanations for why those patterns exist. The hope is to develop propositions regarding what one might be able to expect to see in alternate settings.
2. The companies involved in your study are diverse. Can you outline the types of companies you engaged with, and explain the three-stage interview process that you utilized to capture their responses to the COVID-19 pandemic?
We embarked on this study, motivated by a couple of cases we learned of that involved firms with AM (additive manufacturing) capabilities shifting their production to fill the need for swabs, respiratory components and PPE during COVID-19. We presumed that certain features of their manufacturing experience well-equipped them for this kind of production pivoting, where other firms might find such a shift more challenging and hard to justify. To examine what those features might be, we sought to examine firms across a range of industries, each of which had in-house AM capabilities, but only some of which chose to shift their production to COVID-19 related production. Initially, we only had some suspicions of what features would play a significant role in motivating such pivoting. However, as evidence began to point towards a pattern of GD experience playing a major role, we further distinguished between firms that did or did not have such experience. That was the focus of the later stages of our interview efforts.
3. Your research delves into the application of generative design (GD) tools by companies that opted to pivot their operations through AI. Can you discuss the nature of these GD tools? Do they necessitate special expertise for effective use? Were there observations on how swiftly designers transitioned from being beginners to proficient users of these tools?
To be clear, the pivoting we were examining was not AI-specific. Rather all of the pivoting involved AM (Additive Manufacturing) resources. In fact, most of the work done in that pivoting did not use GD (AI) directly. The designs in question were largely regulatory-specific. Rather it became apparent that the experience engineers had with GD (AI) was critical in structuring their perspectives on the possibility of manufacturing pivoting. That is, experience with GD (AI) increased the openness of engineers to radical changes in what they could produce. It increased their ability and willingness to pivot in even non-GD (AI) contexts.
4. I am interested in the early stages of this project. Your first round of interviews take place in May 2020, just a few months after the start of the pandemic. When did the idea to focus on the impacts of GD as a strategic pivot crystallize for you? When were you first introduced to GD as a significant tool, and how have your perceptions of its utility evolved over time?
This shift really only happened after hearing stories about the general approaches to Additive Manufacturing, from those we were speaking to. When the use of GD came up among those pivoting, even just in passing, it seemed like the kind of pattern worth looking into a bit further.
5. You've made a critical observation in section 4.2.1, stating: "Because aspects of design quality (e.g., strength and durability) are not compromised through GD optimization processes, the more engineers and their managers are exposed to design concepts emerging from the use of GD, the more they are open to new designs that they might now otherwise have come up with themselves." This remark seems to capture an essential aspect of the potential offered by AI-enabled tools. Could you expand on this correlation between 'exposure to design concepts' and 'openness to new designs'? How could this observation apply within the contexts most relevant to my readers, namely AI-enhanced teacher and student work cycles?
Rather than viewing AI as a solution or substitute for critical thinking, it may be more useful to consider how human interaction with AI impacts the way humans think. Humans can learn from interactions with AI, just as AI can learn through human interaction. This symbiosis, and second-level learning, can represent a remarkable shift in the development of how complex thinking can evolve. And it might transcend a wide range of contexts.
6. In section 4.2.4, you reference Na and Kim's assertion that "GD can…reduce friction between designers, developers, and clients. It is expected that it will be able to help establish a new convergence and complex process as an alternative to minimizing and developing competitive products." Throughout your paper, you suggest that GD fosters a distinct form of communication among designers. Could you unpack how GD does and discuss the implications for teacher and student work cycles?
Fundamentally, we need to begin by recognizing that both humans and GD (AI) are going to come up with both terrible ideas and also amazing ideas. And pretty obvious ideas in between. It is the dialogue between the two that help rationalize, vet and highlight the best ideas, and take these to the next level. The use of GD really is a collaborative effort. An exchange of inputs and outputs, a scrutiny of ideas and a recalibration of assumptions to make subsequent dialogues more effective.
7. The concept of generative AI fostering new convergences is fascinating, yet the integration of such tools into educational frameworks is met with a myriad of challenges. From your vantage point, what advice might you offer educators who are navigating the hurdles of convincing various stakeholders of these technologies' value?
Again, as long as all parties recognize both the benefits and pitfalls of solutions generated by AI, and are willing to engage in dialogues with it, learn “with it”, there can be a range of benefits to many stakeholders.
8. In your teaching environment, how do you broach the topic of AI as a "symbiotic learning mechanism" with your students? What strategies have you implemented to ensure that students fully engage with and benefit from these cutting-edge applications?
Great question. Apart from advocating for research at the intersection of Operations Management and technology, in my role as Editor-in-Chief of the Journal of Operations Management, I’m also the co-director of Fisher’s ranked Specialized Master in Business Analytics program. We often get a question from incoming students about whether AI is poised to replace Analysts. Clearly it is not. Like any new tool, it just means a change in what Analysts are responsible for. Simply ceding analysis to a black box, and trusting it (or placing full blame on it) isn’t responsible. The modern analyst’s role is to Scrutinize. Interpret and Question what AI is handing you. If you can’t do that, you have no place doing analysis. Incidentally, we also reinforce the role of sensemaking and scrutiny in frameworks like the OUtCoMES Cycle, discussed in Mastering Project Discovery, a new text on managing analytics projects.
9. Lastly, could you offer a more detailed explanation of the double-loop learning model you've visualized, particularly around the integration of GD? The aspect of 'Strategy, Structure, Decision Rules' within this framework intrigues me; can you explain further. I am also seeking clarity on the operational differences between Loop 2a and Loop 2b. My readers showed great interest in this diagram when I shared it, and I'd like to provide them with a deeper understanding of its components.
In that figure, we can think of Loop 1 as simply trying out a variety of pre-conceived options, one at a time, without any real changes in the way we solve problems. We might learn what works and what doesn’t, but we aren’t learning why. In Loop 2a, we are taking the time to think about why certain things work and others don’t. We are looking for patterns, changing our perceptions of how the world (or the focal context) works. That abductive reasoning is a little like what we went through in this research; adjusting our view of what mattered and why those things mattered, as evidence presented itself. Loop 2b brings in the additional role of the other collaborator: the GD (AI) in this case. We share information with it and allow it to derive ideas based on that information. In our scrutiny of those ideas, we have the opportunity to ask “why did it come up with that idea?”, and “why didn’t I?”. That also has the chance of abductively influencing our mental models of reality. And, in this instance, how our rationalization of AI recommendations performs also provides inputs into future learning. In short, the presence of AI expands the opportunities for double loop learning to occur; If we are willing to take advantage of it thoughtfully.
Takeaways:
Partnership and Balance with AI:
Dr. Bendoly emphasizes the symbiotic relationship between humans and AI, advocating for a partnership where AI complements rather than overshadows human capabilities. He suggests that interaction with AI can change human thinking, enhancing creativity and problem-solving in ways that extend beyond AI's computational advantages.
Learning from AI Interaction:
Through his research, Dr. Bendoly highlights how engineers and designers who work with AI, especially generative design (GD), become more open to exploring new solutions. He notes, "experience with GD (AI) increased the openness of engineers to radical changes," suggesting that AI can broaden perspectives and encourage innovative thinking.
Human Oversight in AI Integration:
Dr. Bendoly discusses the importance of human engagement in evaluating and interpreting AI's contributions, stating, "Simply ceding analysis to a black box, and trusting it (or placing full blame on it) isn’t responsible." This underscores the need for professionals to critically assess AI-generated outputs, ensuring that AI is used as a tool for enhancement rather than a substitute for human judgment.
Adaptation and Evolution of Professional Roles:
Reflecting on the impact of AI on jobs, Dr. Bendoly clarifies that AI does not replace analysts but changes their responsibilities. He argues for a shift in focus towards scrutinizing, interpreting, and questioning AI outputs, highlighting the evolving nature of professional roles in the presence of AI technologies.
Double-loop Learning with AI:
Dr. Bendoly introduces the concept of double-loop learning in the context of AI, explaining how this model facilitates a deeper understanding of why certain approaches work. He describes how AI can contribute to this process by offering new ideas for consideration, which in turn challenges and expands human cognitive models.
What is Abductive Reasoning?
Abductive reasoning is a form of logical inference that starts from an observation or set of observations and then seeks the simplest and most likely explanation.
This process does not guarantee the correctness of the hypothesis; instead, it offers the most plausible explanation based on the available information.
Abductive reasoning contrasts with deductive reasoning, which starts with a general statement or hypothesis and examines the possibilities to reach a specific, logical conclusion, and inductive reasoning, which begins with specific observations and measures to make broad generalizations.
Source: Communication Theory
I want to extend my heartfelt thanks to Dr. Bendoly for sharing his invaluable insights with us. His exploration into the dynamic relationship between humans and AI sheds light on the potential for mutual growth, the necessity of critical engagement, and the ever-evolving nature of our professional roles.
Through his work, he has not only enlightened me but also my readers of Educating AI, encouraging us to approach the complexities of AI integration with a mindset geared towards collaboration, ethical application, and an unwavering commitment to lifelong learning. Dr. Bendoly, your expertise and thoughtful discourse have profoundly impacted our understanding of the future of AI in our lives and work, inspiring a deeper and more nuanced conversation. Thank you for your significant contributions and for guiding us towards a more informed and thoughtful engagement with AI technologies.
See everyone next week!!!
Nick Potkalitsky, Ph.D.
I do enjoy this discussion. Innovation v. creativity is interesting. My guess is that AI has the jump on innovation.The artist's duty is to deliver an emotion to the audience. What I've seen so far with AI are some images that are very pretty but not yet emotive. This is a thrilling time. One of the Substack creators had an interesting take on it. To paraphrase... in the future the importance of content will be replaced by culture.