Mentor in the loop–Is there a role for AI in mentoring relationships?

By Jean Rhodes

In recent years, my colleagues and I have been exploring how web and mobile platforms can enhance mentoring programs—–not replacing face-to-face connections, but enabling efficient between meeting communication and collaboration. Such platforms also allow mentors and programs to share resources, deliver personalized feedback, and track mentees’ progress.  They are also extremely well positioned to integrate machine learning and artificial intelligence (AI) in ways that could further transform the reach and effectiveness of mentoring support. Indeed, there is growing interest in exploring how best to combine mobile technology, AI, and human support. 

So, just as my colleagues and I have suggested that less demanding tasks can be shifted from highly trained professionals to lower-cost supervised mentors, might also consider whether (and how) we should shift certain mentoring tasks to supervised AI. The AI might be able to affordably do what humans cannot always do (e.g., quickly suggesting answers questions, analyzing real-time data, scanning chats for mental health risk, sending rapid alerts to program support). Chatbots could, for example, nudge mentors or their parents to complete intake or evaluation surveys, and answer questions about the program, upcoming events, etc. that can be matched with simple, pre-populated answers.

Of course, AI remains far less adept at navigating more interpersonally complicated communications in ways that are truly aligned with relationships and are limited in their ability to consistently promote the sense of belonging and connection that contribute to mentees’ wellbeing.  They sometimes generate inappropriate, culturally biased content (Bender et al., 2021) and often struggle to maintain the context and relevance over long conversations (Radford et al., 2019) particularly when dealing with ambiguous student inputs (Brown et al., 2020). Since AI responses are primarily based on pre-existing data and patterns, they struggle to understand the broader context of mentees’ lives or changing community norms and contexts. Likewise, because AI models are trained on large datasets that may contain sensitive information, which could lead to privacy issues that require human oversight (Hao, 2020). 

Given these limitation, there have been growing calls for human oversight to address the limitations of current AI models, or  “human in the loop” model to provide oversight and supervision. This would ensures that mentors are there to do what technology cannot do (e.g., problem solve, provide humor, adapt easily, express genuine concern and understanding). Despite the promise of combining AI with human oversight, scaling such oversight in ways that keep pace with the spread of AI remains elusive. Mentoring programs, particularly those using tech platforms, may offer an unusually robust context for  pairing AI-enhanced features with trained and low-cost providers.

In such models, mentors could validate AI-generated recommendations, tailoring them to each mentees’ circumstances and maintaining the authenticity of the responses. Thus, rather than replace mentor support, decision support tools, including AI algorithms, could be used to aid mentors in supporting their mentees. Combining mentoring and AI has could enable mentors to provide targeted, personalized support to a more mentees than they might have been able to offer without AI support, with the potential of  improving programs’ capacity to manage data, provide personalized support and ensure mentee success.  At its heart, successful mentoring is based on understanding individual needs, effective communication, and timely guidance – all areas in which AI can make a significant contribution. The key will be to develop models that enable AI to extend the effectiveness and reach of mentees without undermining the authenticity, safety, and effectiveness of the mentoring relationships (no easy task!). 

Still interested? Below are some of the ways that AI can help college peer mentoring programs, something we are working on with our platform MentorPRO.

Personalized learning: One of the most significant benefits of AI in mentoring is its ability to customize learning for each individual student. Machine learning algorithms can analyze a student’s academic history, learning style, strengths, and weaknesses. This information can then be used to create personalized learning plans, with tailored advice and resources. Mentors can use this information to provide more precise and effective guidance.

Streamlined Communication: AI chatbots can facilitate communication between mentors and mentees. They can manage routine queries and tasks, like scheduling meetings or providing reminders about assignments. This frees up time for mentors to focus on more substantive, nuanced conversations with their mentees. Furthermore, AI chatbots can offer 24/7 support, answering frequently asked questions or directing students to appropriate resources when mentors are not available.

Data-Driven Insights: AI can analyze student data to identify trends, patterns, and correlations that might not be immediately evident. For example, it can identify at-risk students who might be struggling academically or socially, allowing mentors to intervene proactively. This data analysis can also offer insights into which mentoring strategies are most effective, facilitating continuous improvement of the mentoring program.

Skill Development: AI can play a  role in developing key skills for both mentees and mentors. For mentees, AI-driven interactive programs can provide practice and feedback in areas such as critical thinking, problem-solving, and communication. For mentors, AI can provide training scenarios (e.g., ethical and/or cultural issues) that enhance their mentoring skills, and offer feedback on their mentoring approach.

Scalability: With AI, the scalability of mentoring programs can be significantly increased. One mentor can only effectively manage a larger  number of mentees. AI can track the progress of many more students simultaneously, ensuring that each student receives the attention they need. This can be particularly useful in larger institutions where one-to-one mentoring might otherwise be logistically challenging and there is a need for scaling.

Career Guidance: AI algorithms can make use of large datasets about career trajectories and employment trends to provide up-to-date career advice. This can help mentors guide students towards internships, courses, and career paths that align with their interests, skills, and the current job market.

While these are all interesting, we should proceed with caution. There is an urgent need for  research to determine how best to keep the mentoring relationship at the center of this conversation.