Cultivating Organizational Expertise: Mentors or Centaurs?

As someone who spent my early career in IT development and my later career studying collaborative knowledge work, a recent conversation resonated deeply. A manager lamented how some younger IT staff often struggled to find solutions but seemed reluctant to ask questions or get help, endlessly "churning" instead of making progress. "I just want them to find an answer and move forward," she said. We wondered whether their hesitancy to ask for help was an effort to avoid being seen as insufficiently competent. We also realized that in WFH settings, such knowledge-seeking behavior needs to be much more intentional and explicit than it might have been in the past, potentially making it feel even more risky.

The manager's frustration exposed a pervasive challenge: tackling novel problems tied to an organization's complex, idiosyncratic environment of infrastructure, policies, and procedures. While training provides a foundation, truly nuanced problem-solving in this sphere frequently demands deeper organizational familiarity.

The Classic Path: Mentors and Communities of Practice

Traditionally, junior employees turned to seasoned mentors who had cultivated this contextual wisdom over years of embedded, hands-on experience. An illustrative case came from Julian Orr's influential studies of Xerox machine repair technicians in the 1980s [1]. Orr's ethnographic work revealed that although the field technicians were individually responsible for different clients' equipment and ostensibly worked independently, they convened daily in the cafeteria to troubleshoot particularly vexing equipment faults together. Despite reference to official manuals, diagnosing esoteric equipment faults required consulting unofficial "communities of practice" where more experienced technicians?shared hard-won insights.

These organic mentorship networks allowed knowledge to propagate beyond what any individual's training covered. They reflected how expertise arises not just from codified processes but from the living "practice" honed by an organization's embedded specialists over years of real-world problem-solving.

Scholars John Seely Brown and Paul Duguid expanded on Orr's findings in their seminal work contrasting processes and practices, to highlight the importance of contextualized knowledge that formal processes might miss [2].

Brown and Duguid used this contrast to make broader points about organizational learning and innovation. They argued that strictly codifying knowledge into rigid processes is insufficient; true expertise requires cultivating communities where evolving practices capturing nuanced on-the-ground realities can be nurtured and disseminated.

This framing reinforced the value of experienced human mentors guiding junior employees in developing situationally appropriate practices tailored to their organization's idiosyncrasies. Expertise, in this view, transcends one-size-fits-all processes documented in manuals.

Disruption of "water cooler" learning opportunities

For IT staff and other roles embedded within often byzantine organizational environments, such mentorship can be invaluable. Informal "water cooler" discussions used to offer junior employees exposure and access to more experienced peers who could mentor and guide them through an organization's unique technological or other quirks.

The pandemic and ensuing shift to remote work, however, has disrupted this classic mentorship model's viability. With fewer spontaneous in-person interactions, junior employees can struggle to absorb the unwritten "practiced" knowledge of seasoned veterans steeped in their organization's nuances. Isolated at home, they may even lack opportunities to develop collegial relationships that would support them in asking questions more explicitly.

A New Path: The Human-AI Centaur

In this environment, an intriguing alternative has emerged - having junior staff form "human-AI centaur" partnerships. In these symbiotic collaborations, the employee provides high-level framing while an advanced AI language model augments their reasoning and results through iterative dialogue.

Unlike conventional software codified from deterministic processes, these conversational AI assistants can act as "sounding boards" in exploring irregular problems. Less experienced employees can use the AI's ability to rapidly process information across domains, propose fresh perspectives, and engage in substantive back-and-forth exchanges to surface insights encompassing and potentially transcending documented processes. We might even consider that an AI assistant’s broad training data, encompassing the vast online repository of past knowledge-sharing, in many ways represents an unbounded "community of practice" encompassing a rich, evolving body of contextualized "practiced" knowledge.

Interacting with such an AI might present a face-saving avenue for asking questions to improve performance, especially for employees at the lower end of the skill or performance distribution [3], while potentially playing to younger generations' preference for screen-based vs. in-person exchanges.

Mentors, Centaurs or a Hybrid Future?

So which path forward for cultivating organizational expertise proves most effective: the traditional mentor model passed down through communities of practice, or the cutting-edge centaur model fusing human and machine intelligence? Or perhaps an integrated hybrid?

The mentor model ensures solutions stem from deeply grounded experiential wisdom tailored to an organization's unique realities. However, it requires having enough seasoned experts willing and able to effectively mentor.

The centaur model offers potential to augment that expertise at scale through partnership with AI assistants that in many ways have absorbed the "practices" across a vast cross-contextualized corpus. Yet challenges like AI confabulations or blindspots from training limitations remain. There's also still the unknown element of how much real-world context even cutting-edge AI can realistically absorb and reliably reference.

Potentially, a hybrid approach harnessing the strengths of each could emerge. With mentors guiding junior staff working in centaur partnerships, the combination of human and machine intelligence may propel problem-solving forward as organizations navigate constant novelty.

Ultimately though, as Brown and Duguid highlighted, processes alone are inadequate for real-world ambiguity. Any approach must nurture contextualized "practiced" knowledge to truly cultivate expertise. Exploring creative amalgamations of human-centered mentoring and machine augmentation could forge potent new paths for growing institutional knowledge and expertise.

Notes:

1.????? Julian E. Orr’s dissertation “Talking about Machines: An Ethnography of a Modern Job” was published for a wider audience in 1996.

2.????? In their 1991 paper "Organizational Learning and Communities-of-Practice: Towards a unified view of working, learning and innovation," John Seely Brown and Paul Duguid analyzed how the Xerox technicians' community enabled sharing vital contextualized knowledge and practice that formal processes overlooked.

3.????? Multiple recent experimental studies suggest that AI can benefit lesser-skilled or lower-performing users even more than it benefits highly skilled or already high-performing users. For example:

Dell'Acqua, Fabrizio and McFowland III, Edward and Mollick, Ethan R. and Lifshitz-Assaf, Hila and Kellogg, Katherine and Rajendran, Saran and Krayer, Lisa and Candelon, Fran?ois and Lakhani, Karim R., Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality (September 15, 2023). Harvard Business School Technology & Operations Mgt. Unit Working Paper No. 24-013, Available at SSRN: https://ssrn.com/abstract=4573321 or https://dx.doi.org/10.2139/ssrn.4573321