It may be intuitive to think that the tech-savvy young generation, experienced in digital tools including generative AI, would effectively train older employees to use these powerful tools. However, this intuition does not necessarily mean that young people are effective in training their older colleagues on generative AI, according to a new study.
A recent study conducted by Harvard Business School and the Massachusetts Institute of Technology (MIT) challenges the assumption that young professionals can effectively train their older peers to use Generative AI. The study found that younger professionals often suggest inappropriate methods for risk mitigation due to their lack of technical experience and limited exposure. In July and August 2023, researchers surveyed 78 young consultants at Boston Consulting who interacted with GPT-4 as part of this study focused on risk mitigation in generative AI usage. The young consultants proposed strategies to mitigate risks, but the researchers found that the vast majority contradicted the recommendations of generative AI experts.
The researchers noted in their study that interviews revealed two findings that contradicted prior assumptions. The tactics recommended by the younger consultants to address the concerns of their more experienced colleagues differed from those suggested by experts in generative AI. Thus, it became clear that young people might not be the best source of expertise in the effective use of this modern technology. In this context, Dr. Alaa Abdulrazak, an assistant professor at Weill Cornell Medicine in Qatar, mentions in an interview with "Al-Nahar Al-Arabi" that "the tactics recommended by the young consultants to address their older colleagues' concerns about generative AI differ from those proposed by experts in the field."
He adds, "This discrepancy highlights an important fact: that despite their practical experience with technology, these young individuals may not represent the most reliable source for advising their more experienced colleagues on the effective use of this technology." He explains that this result indicates a gap between general knowledge gained from simple usage and deep understanding, suggesting that expertise in generative AI requires more than just a general proficiency and necessitates a theoretical and precise understanding typically possessed by generative AI experts.
Abdulrazak concludes that "these results underscore the necessity of organized training and formal education in generative AI for both young and older consultants. Relying solely on the experimental ideas of younger individuals may lead to suboptimal adoption of technology. Thus, leveraging expert knowledge and establishing comprehensive training programs may close the gap between practical use and theoretical understanding."
In the study, the problems faced by young professionals centered around three main points:
- A lack of deep understanding of the technology.
- A focus on changing human routines rather than considering the broader picture of technological system design.
- A focus on specific procedures related to their work rather than on the environment or technical system.
Dr. Abdulrazak pointed out that the status of older professional generations is not a barrier to receiving training from their younger colleagues. He elaborated: "This does not mean that the recommendations of young professionals are necessarily flawed. If we examine the responses in the study’s appendix, we will see that young professionals possess a foundational understanding of the potential pitfalls of generative AI. However, their risk mitigation recommendations are not sufficiently systematic or overly rely on personal human judgment."
He continues: "This shift in perspective suggests that the resistance of more experienced professionals may be based on practical considerations and not solely linked to rank or ego. These individuals primarily focus on the performance of AI in terms of accuracy, interpretability, and taking context into account."
The study comes at a time when organizations are grappling with the opportunities and challenges posed by current generative AI systems. The study emphasizes understanding not only the inefficiency of youth in training their older colleagues or their narrow vision regarding generative AI but also addressing expert inputs, the overall governance of that technology, and working to improve skills related to organizational work.
Abdulrazak adds, “It is useful to directly address these specific concerns, perhaps through case studies, data-supported guarantees, and demonstrating interpretability benefits among others. Organizations may need to reassess their strategies in implementing new technologies, ensuring a focus on the quality and reliability of outputs. By addressing the practical concerns identified by the younger professionals, companies can facilitate a smooth adoption and optimal utilization of generative AI tools.”
He concludes: “These findings call for a precise understanding of knowledge transfer in workplaces driven by AI. It is not just about bridging the 'status gap' between younger and older individuals, but also addressing practical concerns regarding the effectiveness of technology.”
The researchers in the study recommend that corporate decision-makers should not assume that young people are inherently better at using or even teaching generative AI. Instead, companies should train both young and older employees on the specific risks and requirements of these technologies.
