The Work That Changes: Augmentation and Anxiety

Every AI initiative on campus eventually runs into the same quiet question, asked in break rooms more often than boardrooms: what happens to me?

At the end of an AI workshop for staff, after the demos and the use cases and the reasonably good coffee, a woman who had been quiet the whole session raised her hand. She did not ask about prompts or tools. She asked: “Is this training me to do my job better, or training my replacement?” The room went still in the way rooms do when someone says the thing everyone was thinking.

The facilitator gave the standard answer, augmentation not replacement, and it landed the way it usually lands: politely, and not at all. Not because the answer is false, but because she had no reason yet to believe it. Slogans do not carry that kind of weight. Evidence does, and most institutions have not yet given their people any.

The institutional question

What happens to the people whose work AI changes?

Notice the phrasing. Not “will AI replace jobs,” which invites either denial or doom, both unhelpful. The realistic near-term picture on most campuses is that whole jobs mostly do not disappear, but the composition of nearly every job shifts: the drafting, summarizing, scheduling, and first-pass work compresses, and what remains is the judgment, the relationships, and the exceptions. That shift can be a genuine upgrade to working life or a quiet degradation of it, and which one it becomes is not decided by the technology. It is decided by how the institution handles the transition, which makes this a leadership domain, not an HR afterthought.

What this looks like in practice

Start by taking the anxiety seriously, because it is not irrational. People read the same headlines leadership reads. They notice when the augmentation message is delivered by the same institution that has been managing vacancies through attrition for a decade. Dismissing that as resistance to change is both unkind and inaccurate. The staff member asking “what happens to me” is doing exactly what the institution claims to want: thinking ahead. The first honest move is to say out loud that the anxiety is legitimate, and that the institution’s plan has to earn trust rather than assume it.

Earning it starts with specificity. “Augmentation not replacement” becomes credible the day the institution can say, role by role: here is what AI will likely take off your plate, here is what your role gains, and here is what we will do together if a role genuinely shrinks. Show the advisor that the tool drafts outreach so more hours go to the students who need a human. Show the coordinator what disappears and what deepens. And where the honest answer is that a role will contract, say so early and pair it with a real path, retraining, redeployment, attrition rather than layoffs where possible, because people can handle hard truths far better than they handle pleasant vagueness followed by surprises.

Then there is the training itself, where the sector’s current answer is mostly to hope people figure it out on their own. Self-directed learning favors the people who already have slack, confidence, and curiosity, and quietly leaves behind everyone whose workload or temperament does not allow for after-hours experimentation. That sorting is not neutral. It tends to track existing hierarchies, which means an institution that trains by hope is deepening its inequities while calling it professional development.

The workable structure is not exotic. A foundation for everyone, so the whole institution shares a floor of fluency and nobody has to pretend. Role-specific depth for the people whose daily work is changing most, advisors, enrollment staff, business office, IT. And a supported track for the enthusiasts who will become your internal champions, because peer example moves a campus culture faster than any mandate from the center. What makes any of it work is not the curriculum. It is time. Training bolted onto a full workload is a tax dressed up as a gift, and staff experience it exactly that way. Protected hours are the difference between development and theater.

Faculty need their own version of this, not a staff program with the nouns changed. Their questions are different: what AI means for their discipline, their assessments, their scholarship, and their professional identity, which for many is woven into exactly the kind of intellectual work AI now imitates. The redesign labor from the last issue lands here too, and it is real labor. Fellowships, course releases, discipline-specific communities of practice, and sandboxes to experiment in are what taking that seriously looks like. One more group belongs in this paragraph explicitly: adjunct and part-time faculty, who teach an enormous share of sections and are almost always left out of development programs. An institution whose AI readiness stops at the tenure line has not actually addressed its academic workforce.

Two structural pieces round this out. Someone has to own change management, by name, because a transition this broad does not coordinate itself, and when everyone owns morale, no one does. And the incentives have to point the same direction as the rhetoric: if AI fluency matters, it should show up in job descriptions, evaluations, and visible career paths, so that the person who invested the effort can see where it leads. People watch what gets rewarded far more closely than what gets announced.

The throughline is trust, and it is worth naming plainly: the biggest risk in this domain is not that people learn AI too slowly. It is that the institution spends its credibility on a slogan, and then wonders why adoption is shallow, why the shadow use continues, and why its best people quietly interview elsewhere. Capability follows trust. Rarely the reverse.

The Atlas connection

In Atlas, our AI operating map, this is the AI fluency and professional development domain, and it lives in the Equip group, whose whole question is the one this essay keeps asking: how do we bring our people with us? It is load-bearing for Educate all the same. The teaching redesign and the assessment redesign from the last two issues are, concretely, thousands of hours of human work, and this domain is where those hours either get funded or get wished for. It reaches back into Defense too, because shadow AI shrinks when people are trained on sanctioned tools, and the culture questions here preview Equip’s other half, change, culture, and communications, later in the series. An institution can buy every tool on the market. Its AI capability will still be exactly as deep as its people’s capability, because that is what institutional capability is.

Questions worth putting on the agenda

Honest answers here require HR, academic affairs, and finance in the same room.

• Could we tell each major employee group, specifically, what AI is likely to change about their work in the next two years?
• Is our training structured and on the clock, or are we hoping people learn on their own time and calling it development?
• What is our actual commitment if a role shrinks: retraining, redeployment, attrition, or silence until the budget decides?
• Are adjuncts and part-time staff inside our development plans, or does readiness stop at full-time status?
• Who owns change management for AI here, by name, and does AI fluency visibly count in hiring, evaluation, and advancement?

The bottom line

The woman at the workshop was not asking about software. She was asking whether her institution would be honest with her about her own future, and every campus is answering that question right now, mostly through what it does rather than what it says.

Augmentation is not a message. It is a track record. Institutions get to start building one, or start paying for not having one, beginning now.