The Mentoring Paradox in the Age of AI mentoring

by Jean Rhodes

In “Older and wiser: Rethinking youth mentoring for the 21st Century” (Rhodes, 2020), I explored why the field of mentoring has remained somewhat decoupled from the more rigorous guidelines of prevention science, and granted broad immunity from the consequences of disappointing findings. Although a full explanation is beyond the scope of this post, I placed considerable blame on the emotionally freighted term “mentor.” As programs diversified and moved away from the traditional Big Brothers Big Sisters designation of volunteers as “Bigs” and “Littles,” to “mentors,” and “mentees” (Irby & Boswell, 2016), a semantic sleight of hand occurred that continues to have far-reaching, unintended consequences.  In particular, the emotionally-freighted terminology of “mentor” led to the conflation of formal, often shorter ties, with the more spontaneous, “natural” ties that often develop in families, schools, and community settings. Programs and researchers drew little distinction between the two types of relationships and inspirational stories of young lives transformed by the caring touch of a self-sacrificing teacher or an uncommonly devoted volunteer became the aspirational goal posts for formal programs.

As I wrote “Paradoxically, the harder the field pushed to emulate natural mentors, the more it may have weakened the enterprise. The idea that volunteers should be able to routinely deliver transformative ties placed unrealistic pressure on ordinary matches, zapping them of shared purpose and direction, and subverting the delicate balance between finding intimacy and achieving goals.”

Technology writer Edward Tenner (2018) notes that this phenomenon is common in medicine: “We know that the obsession with childhood hygiene, so popular since the early twentieth century, can weaken the immune system. We know that over-prescription of antibiotics can foster superbugs, that liberal use of opioids can reduce their effectiveness and encourage addiction, and that habitual reliance on sleeping pills can worsen insomnia. Few of us renounce medicine or pharmaceuticals, but we have a new respect for natural equilibria.” 

Since the books’s publication, the automation paradox, threatens to further weaken mentoring.  In deploying AI to make support more widely available, we actually risk swinging the pendulum too far in the opposite direction. Using AI to give more young people support could leave them with less human support, in other words, scaling support could end up weakening it. This is an application of what economist Erik Brynjolfsson (2022) calls the “Turing Trap.” He argues that focusing too much on making AI imitate and replace people can lead us to overlook better ways for AI to help people do things they could not do as well on their own.  A human-at-the-helm approach offers a more promising direction. Programs can use AI to organize information, identify relevant resources, and help mentors prepare for conversations, while keeping mentors responsible for interpreting that information and deciding how to respond.

In mentoring, close relationships are vitally important. But when they are construed as the only active ingredient, we create unrealistic pressure on everyday relationship. But when they are viewed as something that can be replaced by AI, we risk diminishing the very foundation on which all effective mentoring rests. 

References

Brynjolfsson, E. (2022). The Turing trap: The promise & peril of human-like artificial intelligence. Daedalus, 151(2), 272–287. Article DOI

DuBois, D. L., & Keller, T. E. (2017). Investigation of the integration of supports for youth thriving into a community-based mentoring program. Child Development, 88(5), 1480–1491. Article DOI

Irby, B. J., & Boswell, J. (2016). Historical print context of the term, “mentoring.” Mentoring & Tutoring: Partnership in Learning, 24(1), 1–7. Article DOI

Rhodes, J. E. (2020). Older and wiser: New ideas for youth mentoring in the 21st century. Harvard University Press. Book recordabebooks

Tenner, E. (2018). The efficiency paradox: What big data can’t do. Alfred A. Knopf. Publication record