There is a moment that has started to surface in nearly every conversation I have with families preparing for the 2026 and 2027 college admissions cycle. It usually does not begin with a direct question. It shows up more subtly. A pause. A hesitation. A parent glancing at their student’s essay and then back at me, as if asking something they are not quite sure how to articulate.
Eventually, it comes out. “We’re worried this sounds like AI.”
Not in an abstract way. Not as a philosophical concern about technology. It is far more specific than that. They are worried that the essay feels polished but empty. That it reads smoothly but leaves no impression. That it could have been written by anyone.
And what most families do not realize until it is too late is that this instinct is correct, but for a very different reason than they assume. The fear is not that colleges will “catch” AI. The risk is that the application will be quietly compressed into something forgettable.
I have now seen this pattern enough times to recognize it immediately.
In 2025, the conversation around AI in college admissions centered almost entirely on detection. Families were trying to understand whether tools like ChatGPT could be identified, whether essays would be flagged, whether using AI would create risk.
But as we moved into 2026, something more consequential began to take shape behind the scenes. Admissions offices adapted by changing how applications are processed altogether.
What I see repeatedly, particularly at more selective institutions, is that applications are increasingly being run through internal systems that generate high-level summaries before a human reader ever engages with the file. These summaries pull from the Common App essay, supplemental responses, activities, and sometimes even contextual information from recommendations. The goal is to focus human judgement, not replace it.
When an admissions officer has minutes, not hours, to understand a student, the summary becomes the first impression. And that is where strong but generic essays begin to disappear. A student can submit an essay that is grammatically flawless, logically structured, and technically sound, yet when it is distilled into a summary, it reads like this:
A student interested in biology and community service.
No edge. No dimension. No signal.
What most families do not see is how quickly that compression happens. They feel good about the essay because it reads well in full. They do not see what it becomes when reduced. And admissions is, increasingly, a process of reduction.
This is particularly nuanced for students applying within California. The University of California system still relies on human readers for Personal Insight Questions. There is no confirmed AI filtering layer at that level. But that does not mean the bar is lower. In fact, the opposite tends to be true.
When thousands of essays pass through human readers trained to recognize patterns, anything that feels templated is filtered out instinctively. This happens simply because it fails to differentiate and not because it violates any rule.
At the same time, institutions like California Institute of Technology are experimenting with systems such as VIVA, designed to evaluate authenticity, intellectual engagement, and depth of thought. The intent is not narrowly focused on detecting artificial writing. VIVA is used as a tool to surface real thinking.
That distinction matters more than most families realize. The question is no longer whether a student used AI. Now it’s whether the application carries signal. Once you begin to see it this way, a different strategy emerges which is about deliberately creating contrast.
What we implement with our families is something I refer to as the Human Premium Strategy. It reflects what is actually working in this environment.
The first shift is in how AI is used. Families often assume that if AI can generate a strong essay draft, it should be leveraged as much as possible. In practice, what I see repeatedly is that over-reliance creates uniformity. Essays begin to sound interchangeable, even when the topics differ.
Using restraint would be the more effective approach. AI can be useful in the early stages, helping a student organize thoughts or explore structure. But the core of the essay, the lived experience, the reflection, the internal movement, must come directly from the student. When that balance shifts too far toward generated language, the essay loses texture. And texture is what survives compression.
The second shift is toward specificity. This is where I see the most immediate difference between applications that stand out and those that fade. Students are often taught to reflect broadly. To speak about leadership, resilience, growth. But those words, without context, are empty.
What works are the moments. The sentence a coach said that landed differently than expected. The physical sensation before a performance. The overlooked detail that, in hindsight, marked a turning point.
These are not dramatic events. In fact, they are often small. But they are specific. And specificity is something AI struggles to replicate convincingly. It is also what admissions readers remember.
When I review essays with families, there is a noticeable shift when we move from general statements to concrete moments. The essay becomes harder to skim. It demands attention. That is not accidental and is a function of detail.
The third shift is one that many students resist at first, and parents often misunderstand. Vulnerability. Not as a buzzword. Not as a performative gesture. But as an accurate representation of how a student thinks and changes over time.
What I see repeatedly is that students present outcomes. They describe what they achieved, what they learned, how they improved. What is missing is the tension that led to that change.
Where they were uncertain. Where something did not work. Where they had to adjust their understanding. In a landscape filled with polished narratives, it is the unpolished moments that create credibility; because they are real.
This is the layer that systems like VIVA are designed to surface, and it is the layer human readers respond to instinctively.
There is a simple way to assess whether an essay is at risk. When I ask a student to identify three moments in their essay that only they could have experienced, there is often a pause. If those moments are not immediately clear, the essay is likely too general. And if it is too general, it will not survive summarization in a meaningful way.
Families often search for answers in the wrong place. They look up how colleges detect AI essays, how to avoid being flagged, how to ensure compliance. Those are surface-level questions.
The deeper question, the one that determines outcomes, is whether the application carries enough specificity, voice, and authenticity to resist being flattened. That is where admissions decisions are increasingly being made. And it is also where most families do not realize they are exposed until decisions arrive.
By the time a student is placed on a waitlist with an application that “looked strong,” the opportunity to adjust has already passed. This is why the work has to happen earlier, and more intentionally.
Because the students who stand out in 2026 are not the ones who avoided AI. They are the ones who remained unmistakably human in a system that is designed to filter everything else.