Dave Marshall

The through-line

Marshall's consistent argument is that technology in higher education earns its keep by making people more human, not less, and that it arrives more slowly than the hype cycle claims. Having watched the web move from novelty to infrastructure in student recruitment, he predicted early in the generative AI wave that things would not change overnight, and he holds to that reading: meaningful change in higher education is incremental, and no one in admissions lost their job to a robot processing applications [1]. What has shifted in his account is the temperature of the conversation rather than the substance of his position. Where there was hysteria, "the atmosphere is much calmer now," and institutions have moved to "oh there's these exciting new new tools that we can use let's be thoughtful about how we can use them" [1].

Underneath the caution sits a genuinely large claim. Marshall believes AI is "the most um gosh transformational gamechanging um disruptive if you will technology" for advancing higher education's mission of making education accessible to everybody, "maybe more than the internet itself," and he is careful to flag that he knows these are bold statements he intends to back up [1].

On why change in higher ed is incremental

Marshall's long view is that the problems persist even as the tools change. The same institutional struggles he encountered when websites were text based, presenting the institution well, integrating student engagement with the CRM, communicating with prospective students, parents, current students and alumni, are still live [1]. That history informs his refusal to treat any new technology as a rupture. He also notes how slippery the vocabulary is: "the term AI is just so so broad um we could say we've been doing AI for 20 years now you could say AI didn't really come out until chat GPT showed the world that it was here" [1].

On AI as a commodity, not a moonshot

What institutions have actually found appealing, in his telling, is a chatbot trained to be an expert on their own institution, on the nuances of financial aid, and on the differing situations of transfer, adult and online, and undergraduate students, responding accurately and in whatever language the student types in [1]. He is emphatic that this is not a boast about proprietary capability: "this is not a flex or a brag from this this this is a commodity now" [1]. The excitement, for him, is precisely in the accessibility. "You don't need data scientists and a whole team of people to do this you need to be thoughtful about about how you do it" [1]. The payoff is friction reduction, both for students and parents learning about institutions and for students navigating their experience once enrolled, and it removes the dependency on whether staff happen to be available and on the disruption caused by staff turnover [1].

On preserving staff time for what humans do best

Marshall frames automation as a retention strategy for people, not a replacement for them. "Preserving staff time for what humans do best is the name of the game," he argues, adding that "Staff burn is real" and that humans need connection, food and sleep to be effective communicators [1]. He reports never having met a client or prospective client whose staff saw this as anything but welcome, because it lets them keep their time for genuine, empathetic conversations while the technology handles the repetitive load [1]. The consequences he attaches to that are concrete: "that means less turnover that means better conversations that means more empathy" [1].

On technology that makes people more human

The irony he keeps returning to is that mediated communication can be more candid than face to face. In a text conversation with a friend or family member, "you might even say things that you would never say or you wouldn't you might have a little anxiety if you were right in front of them," which he sees as creating "such a beautiful safe environment to have vulnerable conversations" [1]. His illustration is a large four-year public research institution whose orientation team hired a student worker to text prospective, then accepted, then enrolled students over months. When that student was introduced on stage at orientation, the room stood and cheered, because families had built a real connection with him during what Marshall calls "a vulnerable part of their life" full of questions about whether they will have friends and be accepted [1]. What that student worker achieved, in his phrase, was "empathy at at scale," which he treats as the central opportunity at a moment defined by a lack of connection and by how complicated higher education can be to navigate [1]. He is explicit that this is a step past the old model of personalization, where a communication plan could be checked off with "dear first name" emails and turnout letters [1].

On fitting in versus belonging

Marshall draws a sharp line between the two. Fitting in is "top of the funnel": seeing photos of students who look like you, reading student spotlights, finding programs, and hypothesising that you might fit [1]. He treats that as a necessary step and warns against designing it for a subset of the population, because doing so ignores whole other groups [1]. Belonging is a different order of thing. "It's a whole another energy," no longer a guess: "you feel it and you feel it viscerally and um you can't fake that you know that's that's that's the outcome of connection" [1]. His stated stakes for getting this right at scale run outward from the individual: better for the student, better for the institution, and ultimately better for society, because more students gain access to education, navigate its challenges successfully, and have a better chance afterwards [1]. He has also taken up the question of how texting reaches specific student populations, discussing texting and Black men in higher education with Dr. Walter Kimbrough [2].

Takeaways

• Change in higher education is incremental, not revolutionary; the early panic that AI would displace admissions staff has not materialised, and the atmosphere around adoption is now calmer and more thoughtful [1].
• The institution-trained chatbot, expert on financial aid nuance and on the differences between transfer, adult and online, and undergraduate students, and fluent in whatever language a student types in, is now "a commodity," achievable without data scientists if you are thoughtful about it [1].
• Automation should be justified by staff retention and quality of human contact: "preserving staff time for what humans do best is the name of the game," which yields less turnover, better conversations and more empathy [1].
• Text-based communication can be more candid than in-person contact, creating "a beautiful safe environment to have vulnerable conversations" during a high-stakes moment in a student's life [1].
• Fitting in is a top-of-funnel hypothesis a student forms from photos, spotlights and programs; belonging is visceral, cannot be faked, and is "the outcome of connection" [1].
• Design top-of-funnel materials so the whole prospective population can imagine fitting in, not just a subset [1].
• Marshall argues AI may be more transformational for the access mission than the internet itself, precisely because it reduces friction for students and parents trying to understand institutions [1].