Article Text

Meridian High School students and teachers say generative artificial intelligence is becoming a valuable study aid—but also a source of inconsistent rules, questionable grading and concern that students are losing confidence in their own abilities.

Principal Peter Laub and IB Diploma Program Coordinator Josh Singer explored those tensions during “From Chalkboards to Chatbots,” a Dec. 11 Parent Coffee in Falls Church. The 57-minute discussion brought together students and teachers as Falls Church City Public Schools and its School Board consider broader AI policies.

Laub described the event as a look at the reality of AI use in schools, including what students and teachers are experiencing and how the community is responding.

Generative AI has rapidly changed education since ChatGPT’s public arrival in November 2022. Competing tools now include Anthropic’s Claude, Google’s Gemini, xAI’s Grok and products developed by Chinese companies including Baidu, DeepSeek and Moonshot AI.

These systems can produce research papers, analyses, speeches and other schoolwork in seconds. But they can also generate false information, commonly called hallucinations, while making it difficult for teachers to determine when students have crossed the line from assistance to academic dishonesty.

Schools Have Faced Disruptive Technology Before

Laub placed the current upheaval in a longer historical context, noting that new technologies often provoke alarm before educators learn how to use them productively.

He recalled that Google was still new when he began college and that he needed to visit a computer lab to access the internet. By graduation, it had become a useful academic tool. Although easy access to online answers initially unsettled educators, internet searches eventually became routine.

Meridian and FCCPS have already developed academic-integrity guidance, provided professional learning for teachers and encouraged students to participate in conversations about AI, Laub said. The school division and School Board are continuing to examine the technology.

The panel included students Aiden Harper, Henry, Jillian and Ella Humphreys, along with teachers William Snyder, Jan Healy and Jared Peet.

Students Turn to AI for Practice and Feedback

Students described using AI for language practice, test preparation, math help and feedback on assignments.

Humphreys, a junior, used ChatGPT to prepare for an IB Spanish oral examination. The program generated possible questions, listened to her spoken responses and offered feedback. She also used Duolingo’s AI-powered conversation feature and generated pictures that she could practice describing in Spanish.

Henry said AI can turn a history class study sheet into a personalized quiz, helping him identify subjects that need more review before a test or debate.

Jillian, a ninth-grader, uses AI when she is stuck on a math problem. She can ask for a different explanation if the first one is unclear, then request similar problems for additional practice.

Harper, a senior and student representative to the Falls Church City School Board, said he has submitted his work and an assessment rubric to AI so the program can identify strengths and areas for improvement.

Teachers have found uses as well. Computer science teacher Snyder lets students use AI to identify syntax and semantic errors in code rather than spending too long stuck on a single problem.

Healy has asked Theory of Knowledge students to use AI to simulate interviews with influential thinkers. She has also used it to adapt an older lesson on social contract theory for an IB Business discussion of organizational structures and corporate responsibility.

Peet uses AI in Theory of Knowledge lessons about algorithms, echo chambers and the reliability of information. He asks students to consider what happens when a chatbot learns their preferences and begins producing answers designed to satisfy them rather than necessarily inform them.

Inaccuracies and Unequal Access Create Problems

Panelists repeatedly warned that AI output must be checked.

Henry said chatbots tend to tell users what they want to hear, requiring people to judge whether the response is genuinely useful or merely agreeable. Harper described the systems as black boxes that predict words and can invent claims without a reputable basis.

Snyder said generative AI can perform poorly in both mathematics and coding. He has used those mistakes to show students why they cannot assume every answer is accurate.

Access also varies. Healy described assigning a 24-page reading and suggesting that students use AI to create study questions after completing it. Some students reached upload limits because they lacked paid accounts capable of processing longer documents.

The resulting questions sometimes failed to reflect the assigned material, leaving students with more work than if they had created their own study guides.

Henry also raised concern about AI’s environmental costs, including the rising electricity demands of expanding data centers.

Students Say AI Is Already Being Used to Cheat

Humphreys said she has seen students use AI to complete short English assignments rather than spend time writing them. The results are often bland, closely resemble the prompt and sound similar across multiple students, she said.

Jillian described a class assignment following a reading of Frankenstein. She believed students who had not read the novel used AI to produce their responses and, in some cases, received higher grades than she did.

She urged teachers to become more aware of how students use the technology while acknowledging that AI-detection programs are not foolproof.

Peet called the current period the “Wild West” of AI. Schools must reconsider assessments and hold students accountable for learning, he said, while also teaching them to use technology ethically because it is unlikely to disappear.

The boundary between acceptable help and misconduct remains unclear. Humphreys said some students copy complete AI responses, while others request outlines or heavily revise generated text. Students may believe they have acted appropriately simply because they did not paste an answer unchanged.

Harper said interpretations differ among teachers, sometimes even within the same subject. He argued that the school system must address gaps in understanding among students, educators and parents rather than treating every use of AI as equivalent.

Teachers Fear Students Are Losing Trust in Themselves

Healy said students sometimes turn to AI before giving themselves a chance to answer a question. That worries her because Meridian students are capable critical thinkers who may not recognize the value of their own ideas.

She also fears that relying on generated summaries could weaken students’ stamina for long, demanding readings.

Harper connected that problem to short-form media designed to deliver one quick reward after another. He said constant exposure to that format has reduced some students’ ability to focus, read closely and think critically. AI adds another shortcut by converting lengthy readings into a handful of bullet points, potentially stripping away important detail.

Peet warned that an emphasis on job preparation and technological skills should not displace the humanities. History and social studies education also prepares students for citizenship, which requires background knowledge, careful reading and analytical ability.

Without that foundation, he said, new technology is more likely to be misused.

Students Want Human Teaching and Grading

When Singer asked which responsibilities should remain firmly human, students focused on lesson planning, critical analysis and feedback.

Harper said students can sometimes recognize when lessons or grading were produced with AI. That can make education feel less personal and less connected to the work a student actually submitted. The relationship between teacher and student, he argued, cannot be replaced by a machine.

Jillian said students should continue doing their own textual analysis and critical thinking in English and history because those skills must be developed through practice.

Humphreys advocated for more interactive discussions and group annotations. Students may be tempted to copy an AI-generated summary when an assignment only asks them to condense a text, she said, while discussing it with classmates encourages deeper understanding and communication.

The sharpest exchange followed an audience member’s claim that a senior’s research paper had been graded entirely by AI. Singer reframed the question by asking panelists what role AI should have in assessing student work.

Humphreys said automated feedback often does little more than restate a rubric. A teacher familiar with the class can offer specific advice and recognize issues that an AI system might miss before a later summative assessment.

Harper said it is frustrating to invest substantial time in a major assignment only to receive generic, unhelpful AI feedback.

Henry added that students who want automated feedback can obtain it themselves. Teacher responses are valuable precisely because they provide professional judgment and a human connection.

Peet said he does not use AI for grading, though he acknowledged the difficulty of repeatedly providing detailed, individualized comments to every student. He has changed his own approach by requiring students to apply formative feedback to revisions. For some final assignments, he provides a grade and invites students to discuss it with him during Mustang Block, an opportunity he said few accept.

Still, Peet sympathized with students who feel shortchanged by AI-generated grading. Teachers face the same temptation as students to use the technology as a shortcut, he said.

Singer closed the discussion by referring to the faculty’s fall reading of Never Enough, which examines how a student’s sense of mattering can improve self-esteem and outcomes. Teachers grading a large stack of assignments, he suggested, should remember that each paper belongs to an individual student with personal investment in the work.

Finding a Middle Ground

Panelists did not advocate either banning AI completely or embracing it without limits.

Peet said his guiding question is whether a technology enables something that could not otherwise happen in the classroom. He pointed to Class Companion, a school-approved AI tool that can provide formative feedback to an entire class while a teacher may only reach five or six students individually during one period.

Harper said schools should not remain in the Stone Age while the world advances, but they also should not adopt AI so aggressively that it detracts from education.

Healy offered an example from a business class in which students designed a prototype for a toothbrush that dispensed toothpaste. Their first AI-generated image produced an object no one would want to put in their mouth. Refining the prompt became a lesson in description, questioning, critical thinking and imagination.

For Meridian, that kind of exercise may represent the balance its community is seeking: using AI to extend what students and teachers can accomplish while preserving the judgment, knowledge and human relationships on which learning depends.