Ernil Sabaj

I'm constantly surprised by the paranoia of a certain segment of the academic community when it comes to AI and LLMs being used to write text. I understand the fear of evil AI overlords turning earth into the Matrix and us into batteries — that's a separate conversation. But the hyperventilation about AI being used to communicate scientific ideas is puzzling to me. We are not in the business of writing poetry. We are trying to resolve unanswered questions, accumulate knowledge, explain mysteries that remain unexplained, cure cancer. If a researcher finds that an LLM helps them communicate their results more efficiently — results that deepen our understanding of the universe in any field — then what, exactly, is the problem? The substance is what matters, not the tool used to polish the prose. I am still narcissistic enough to prefer my own text to what an LLM produces. But I have no philosophical objection to using one, and I don't see why anyone else should either. One serious concern is pedagogy. Writing is a thinking tool. Struggling to put an idea into words forces you to sharpen the idea itself. If students outsource that process entirely, they may not learn the cognitive discipline we are trying to teach. But this is hardly a new problem, nor hard to solve. We have known the solution for thousands of years: An exam. Paper and pen in a controlled environment. Oral examination. Socratic dialogue. What is surely a losing battle is policing students with unreliable commercial "detection tools," creating an atmosphere of suspicion and paranoia, and pretending we can preserve a pre-AI world. It's lazy. There is no going back. AI will only get better, and our students' success later in life may largely depend on their ability to use it. The question is not whether they will use it, but if we adapt our teaching to ensure genuine learning — both in the traditional sense and in mastering this new power.