College’s real edge is social capital, not chatbots
In a new study, Emily Hersch and colleagues report that students who completed “Connected Scholars,” a semester-long social capital course at UMass Boston, were dramatically more likely to stay enrolled and earn their degrees on time than similar students who did not take the class, even after accounting for many background differences. At a moment when colleges are racing to roll out AI chatbots as the front line of student support, these findings are a reminder that relationships, not algorithms, remain colleges’ most powerful asset.
Connected Scholars is a one-credit elective designed to teach undergraduates, many from historically marginalized backgrounds, how to recognize the importance of social capital, identify potential mentors, and practice concrete help-seeking and networking skills with faculty, staff, and professionals. The course includes activities like mapping existing networks, role-playing office hours, and attending supervised networking events, with an explicit focus on helping first-generation, low-income, and racially minoritized students navigate the “hidden curriculum” of college and build webs of support on and off campus. Importantly, CS does not assign mentors; instead, it equips students to recruit and sustain multiple mentoring relationships for themselves.
The new study followed more than 30,000 undergraduates over several years, comparing those who enrolled in Connected Scholars (about 900 students) to all other students using university records. Because students chose whether to take the course, the researchers used propensity score matching and multivariate models to adjust for differences such as race, first-generation status, Pell eligibility, high school GPA, and prior academic preparation, an approach that cannot fully rule out selection bias but substantially tightens the comparison. Even with these caveats, the patterns are striking: students who passed Connected Scholars had roughly 1.9 times the odds of being retained after two years, more than three times the odds of graduating within four years, and nearly three times the odds of graduating within six years compared with all other undergraduates, while showing no meaningful difference in cumulative GPA at graduation.
Two additional findings make the results especially important for equity. First, the students who opted into Connected Scholars were, on average, more vulnerable to attrition than their peers: they were more likely to be first-generation, Pell-eligible, Black, and to have slightly lower high school GPAs. Second, only students who actually passed the course showed these large boosts in retention and graduation; students who enrolled but did not pass looked much more like non-participants on most outcomes, suggesting that exposure to the full curriculum, not mere interest, is doing the work. In other words, a low-cost, one-credit class that explicitly teaches students how to build social capital appears to bend long-term outcomes in ways that traditional advising and “study skills” courses often do not.
These findings arrive just as colleges are embracing a very different kind of solution to the challenges of student support: AI chatbots that promise 24/7 answers at scale. Systems like ChatGPT Edu and campus-branded bots can be genuinely helpful for handling routine questions about forms, deadlines, and policies, and early studies show that automated nudges can nudge up completion of short-term academic tasks. But if institutions start treating these tools as substitutes for human relationships rather than supports for them, they risk undercutting the very social capital that Connected Scholars and similar interventions work so hard to build. Chatbots can remind a student about the add/drop deadline; they cannot walk that student down the hall to meet a professor, or later pick up the phone to recommend them for a job.
The promise of higher education has never been “ask us anything, anytime.” It has always been “come study here, and you will leave with people in your corner.” In an era when information is so easily accessible and answers are instant, colleges that remember this, and design their AI strategies around social capital rather than convenience, will be the ones that deliver on that promise.
Breaking news: Working with single-session intervention pioneers, Jessical Schleider and Mallory Dobias, we recently developed a free, single-session “flash” version of the course, Connected Futures, in which we focus on a key active ingredient (help-seeking). We’re in the early stages of evaluating it and would love your feedback!
