Automate the Mundane to Protect the Human
The machines were never the threat. Forgetting what they are for is.
An article from the Build series. Part One of two.
Bring up AI in almost any room (a faculty lounge, a board meeting, a dinner table) and someone will eventually ask the question with a half smile: "So, is this Skynet?"
I understand the joke. Many of us grew up on The Terminator, where the machine wakes up, decides humanity is the problem, and takes over.
The Doom Story Has Real Narrators
It would be easy to wave the Skynet question away. I don't think we should, because some of the people closest to this technology are telling a version of that story in earnest.
Geoffrey Hinton is the clearest example. He is often called the godfather of AI, and he shared the 2024 Nobel Prize in Physics for the neural network research that made today's tools possible. In 2023 he left Google so he could speak freely about the risks, and he has since put the chance that AI leads to human extinction within the next few decades at roughly 10 to 20 percent. When the person who helped build the foundation says that, the joke stops being only a joke.
Not everyone agrees. Princeton's Arvind Narayanan and Sayash Kapoor argue that AI is best understood as a "normal technology," one that will reshape work the way electricity did: over decades, through slow institutional adoption, not a sudden takeover.
Behind these voices sits a wide field of research institutes, and they do not agree with one another. The Center for AI Safety warns that extinction risk from AI belongs alongside pandemics and nuclear war. Stanford's Institute for Human-Centered AI tracks what is actually happening, year by year, in its AI Index. The AI Now Institute argues that the real dangers are already here: concentrated corporate power, surveillance, and pressure on workers.
I take Hinton seriously. But notice where the warning lands for a school. The extinction debate will be argued in labs and legislatures, far from any admissions office or business office. What reaches a school is quieter and more immediate: whether students still do the hard thinking that builds judgment, and whether adults still do the human work that families are paying for. Those are not questions about the machine. They are questions about us.
So the useful question for anyone who leads a school, a business office, or a family is quieter than Skynet. What is this tool for, and what should it never touch?
I am answering that question in two parts. Part One, this piece, is about the principle and the people: why the line between the mundane and the human matters, what I have seen in my own work, and the simple tools I use to hold that line. Part Two, coming next, turns to institutions: when AI should enter a student's life, what it can take away when it arrives too early, and how a school turns good instincts into policy.
Where I See It Working
Part of my work at BLAC is resume and career-narrative support: helping people tell their professional story on a resume, on LinkedIn, and in the interview room, as they move from one chapter of a career to the next. That work is where I first saw that ignoring AI is no longer a choice anyone gets to make, and it is where I see this approach working every week.
Here is what many job seekers do not know. Before a person reads their resume, a machine often does. In a 2026 Resume Genius survey of 1,000 U.S. hiring managers, 79 percent said their company uses AI somewhere in hiring, and more than a third use it to screen or rank resumes. Recruiters load the requirements from the job description, and the system checks each application against that list. Meet the requirements and you move forward. Miss them, and you can be set aside before anyone learns your name.
So my advice now surprises people who expect a warning: if you want to get past the AI on the other side of the table, you should be using AI too. Use it to read the job description closely, map every requirement to your experience, and make sure nothing the employer asked for is missing. That is mundane, checklist work, and a machine does it well.
But the same survey has a second half. Eighty percent of those hiring managers said they can often tell when a resume was written by AI, and 72 percent said heavy reliance on it makes a candidate seem less skilled. The checklist gets you through the door. Your story, in your own voice, is what makes someone want to meet you. No tool can tell that story for you, because no tool lived it.
The people I support with their resumes are often blown away when they see how this works. Once they understand what is on the other side of the table, they realize that beating AI with AI is simply the cost of being read. The ones who are willing to use it get their resume submitted sooner, and they get something more valuable back: time. Time to research the organization. Time to practice their answers out loud. Time to walk into the interview prepared to tell their story instead of still polishing the document that got them there.
I have tested this on myself, too. In my own job transition, I use AI to check every application against the employer's stated requirements before I submit it, and I am tracking my results. It is too early to put a number on it. What I can already say is that the work has not disappeared. It has multiplied on both sides of the table. Recruiters are reading more applications than ever, and I am creating more tailored documents than ever. AI is what makes that pace possible without losing the substance. It handles the matching, so I can spend my energy on the part only I can bring: the conversation, once I get to have it.
That is the whole idea in one job search.
Let the machine handle the matching. Keep the meaning human.
Schools are facing the same choice now, and they have faced versions of it before.
We Have Been Here Before
The phrase "artificial intelligence" was coined for a summer workshop at Dartmouth College in 1956, and schools have lived through smaller versions of this moment ever since. When graphing calculators like the TI-83 arrived, teachers worried students would stop learning math. When iPads rolled into classrooms, the worry was that screens would replace teaching. Some of those fears were partly right. But most of the damage came not from the tool itself. It came from adopting the tool without deciding what it was for.
What Surprised Me
When I started using AI seriously, four things surprised me.
It gave time back. I expected the fear to be about replacement. Instead, the first thing it did was return hours to me, and the real question became what I would do with them.
It was most useful when it pushed back. The value was never in having something agree with me. It was in having something ask, "What evidence contradicts this?" before I committed to a conclusion.
It exposed my own mess. Before AI could help with a process, it showed me where the process was broken.
Bad workflows do not get better when you speed them up. They get worse faster.
The guardrails mattered more than the tool. The rules I set about what AI could and could not do shaped the results far more than which product I chose.
What the Field Is Already Saying
I wanted to know whether this was only my experience, so I looked at what schools are actually doing. Three examples stood out.
The Lawrenceville School, a grades 9-12 boarding and day school in New Jersey, says plainly that teaching will remain "fundamentally in-person and relational." It built three councils to guide its approach to AI, one for students, one for faculty and staff, and one for outside advisors, and it steers students toward trusted adults rather than AI companions.
Lakeside School, a grades 5-12 day school in Seattle, draws the line in its family handbook: using AI to instantly complete work meant to build a skill is skipping the learning itself. Then it went further. Student tech interns in the school's tech support team, known on campus as the Grotto, built the school's own AI system. Two students realized how simple it was to host AI models themselves. The project started on a staff member's personal gaming computer, and the interns turned it into an in-house server called Andromeda that gives students and teachers access to models like ChatGPT and Meta's Llama. Because the school pays only for what is used, saving more than $100,000 a year by one estimate, access no longer depends on what a family can afford. And because teachers help build its chatbots, including a math bot that gives hints instead of answers, the guardrail lives inside the tool.
Pine Crest School, a PreK-12 school in South Florida, shows the same thinking in the business office. Its CFO, profiled by NBOA, the national association for independent school business officers, automated repetitive tasks and freed roughly two weeks a month for her staff. That time went to growth and succession planning, not cuts. Her first rule: fix a process before you automate it. At The Meadowbrook School, a junior kindergarten through grade 8 school in Weston, Massachusetts, the chief innovation officer keeps what he calls a "human filter": every AI output is a starting point, never the final answer.
Artesha Moore of Association Forum put it in a single line when she spoke to NBOA's staff this summer: "Speed without wisdom is chaos, and wisdom without tools is not enough."
The Pattern Underneath
None of these schools started with the tool. Each decided first what had to stay human: at Lawrenceville, the relational classroom; at Lakeside, the struggle that builds a mind and equal access to the tools; at Pine Crest, the growth of its people. Then each designed around that decision. That is the difference between adopting a technology and building with one.
Schools that struggle tend to do it in reverse: the tool first, a student policy second, and the adult side last, if ever.
The New Hidden Curriculum
There is one more thing the field conversation tends to miss, and it is the part closest to my own work.
The question at the center of everything I do is how people learn to navigate environments that were not built for them. I have spent years listening to people answer that question, in more than forty recorded interviews and 180 published alumni accounts, and the same thing keeps surfacing: unwritten rules. Who you could ask. What was really expected. Which door was actually open. Nobody handed those rules out. Some students learned them at home. Others learned them late, or never.
AI is becoming the next set of unwritten rules. In most schools, the line between acceptable and unacceptable use moves from class to class and assignment to assignment, often for good reasons. But someone has to decode it. Some students have a parent who uses these tools at work and coaches them at the kitchen table on how to use AI well: to brainstorm, to check their reasoning, to ask a better question. Other students hear only "don't use it," or worry that asking about AI will make a teacher suspect them of cheating. Both groups sit in the same classroom. Only one of them is learning the skill their future employers will expect. The same is true after graduation: a job seeker who does not know that software reads their resume first is playing by rules no one told them about.
That is why Lakeside's Andromeda matters, and also why it is not enough on its own.
Access opens the door. Navigation is everything after.
Giving every student the same tool solves the first problem. Teaching every student how to use it well, out loud and on purpose, solves the second. Without that, AI becomes one more advantage that quietly follows the families who already know the rules.
The adults who close that gap are the same people who have always closed it: the teacher, coach, advisor, or staff member who stands in the gap and translates what the institution expects. For AI, that translation has to be explicit. It cannot be left to whoever happens to have the right parent at home.
What I Finally Understood
What I finally understood is that you have to decide what is sacred before you decide what to automate.
And here is why the order matters so much: AI makes an institution more of what it already is. A school with clear values and healthy processes becomes clearer and healthier, faster. A school with a broken workflow and an unexamined culture gets a faster broken workflow and a louder version of the same confusion. A school where some families already know the unwritten rules and others do not will watch that gap grow.
The machine does not decide what your school is. You do.
It simply amplifies the decision, or the absence of one.
What I Built: Five Guardrails
This series asks one question: how do you build something that keeps creating value after you are no longer the one holding it together? For my own practice, including the resume support I described above, the answer was not a tool. It was a set of rules the tools have to follow, written down so they hold on my busiest week, not just my best one. None of it requires a large budget or a technical team.
One source of truth. Every client, conversation, and commitment lives in one system of record. AI works inside it and never creates a parallel version of reality.
AI proposes, I decide. AI can draft, sort, summarize, and flag. It never makes a consequential judgment, and it never silently overwrites anything that matters.
The source stays the source. A summary never replaces the original conversation, document, or data. When it matters, I go back to the real thing.
Every finding faces its counter-evidence. Before anything goes into a report, I ask what would prove it wrong, and anything I publish needs more than one source.
Time saved has a destination. Before I automate something, I decide where the returned time is going. If I cannot name it, the time disappears into more busywork.
You can see these guardrails at work in a single job search. AI does the matching against the job description. The person and I decide what their story is. Their own experience stays the source, and nothing goes out that they have not read and owned. And the time we save goes straight into preparing for the interview, where the human part finally gets its turn.
The Mundane or Human Sort
Any leader can run this on their own week. Take the tasks on your calendar and to-do list and sort each one with four questions.
Does this task require judgment about a specific person? If yes, it stays human. Admissions decisions, a hard conversation with a family, a performance review, an advising moment.
Would anyone feel less cared for if they knew a machine did it? If yes, it stays human, or AI only prepares the ground so the human part can be better.
Is the person doing this task still learning how to do it? If yes, they do it themselves first. A student, a new teacher, or a first-year analyst should master the work before a machine is allowed to do it for them.
Is it repetitive, rule-based, and recoverable if it goes wrong? If yes, it is a candidate to automate. Reconciliations, first drafts, scheduling, meeting notes, formatting, first-pass research.
Then do the step most people skip: for every task you move to the automate column, write down where the time is going instead. That last column is the whole point.
So, is this Skynet? Not if you decide, before the machine does, what it is for.
Guardrails work for one person. A school needs more than that. In Part Two, Skill Before Shortcut, I turn to what an institution has to build: when AI should enter a student's life, what it can take away when it arrives too early, and how to turn guardrails into policy.
If This Sounds Familiar
If your school is using AI faster than it has decided what AI is for, you are not behind; you are at the decision point. I would welcome a conversation about where the human work lives in your institution, and how to make sure every student learns the rules for these tools, not just the ones who learn them at home.
References
AI Now Institute. (n.d.). AI Now Institute. https://ainowinstitute.org
Center for AI Safety. (2023). Statement on AI risk. https://www.safe.ai/work/statement-on-ai-risk
Lakeside School. (2025). Family handbook 2025-2026: Use of generative AI.
Narayanan, A., & Kapoor, S. (2025). AI as normal technology. Knight First Amendment Institute at Columbia University.
National Business Officers Association. (2026, August 31). Your AI questions, answered: Insights from the Byte-Size AI series. Net Assets.
Resume Genius. (2026). Hiring insights report [Survey of 1,000 U.S. hiring managers conducted by Pollfish].
Shields, J. (2026, July 14). A midsummer reflection on AI and generational difference. Net Assets.
Shields, J., & Greene, N. (2026, April 21). Seizing the AI opportunity. Net Assets.
Stanford Institute for Human-Centered Artificial Intelligence. (2026). AI Index report 2026. Stanford University.
The Lawrenceville School. (2026). AI at Lawrenceville.
The Tatler. (n.d.). AI enters the classroom: Grotto interns pioneer Lakeside's first AI. The Tatler [Lakeside School student newspaper].
Notes on sources: Geoffrey Hinton's risk estimate comes from his public interviews following his 2024 Nobel Prize. The 1956 Dartmouth Summer Research Project on Artificial Intelligence is widely documented as the origin of the field's name. Interview and alumni account figures are from BLAC's own archive, gathered since 2021.
Comments