ATLANTA, Aug. 2 (Nationwide Times) — The spread of generative artificial intelligence is pushing educators to rethink how they teach and test students. The central question is no longer only whether a student used AI to complete an assignment. It is whether the student learned enough to explain, evaluate and apply the work without it.
Amy Bruckman, a Regents’ Professor at the Georgia Institute of Technology and senior associate chair in the School of Interactive Computing, said universities should teach students how to use AI while also protecting the skills that students need to develop themselves.
“I don’t think cheating is the right concept to use here,” Bruckman told Fox News Digital. She said some students are using AI to avoid learning, making it harder for instructors to know what students understand.
Why assessment is changing
Traditional take-home essays and other assignments completed outside class can be difficult to evaluate when students have access to systems that generate text, solve problems or summarize readings. An assignment may look polished while revealing little about the student’s own reasoning.
Bruckman said she uses closed-book, paper tests to check what students know. These tests require students to recall and explain course material without an AI tool or other outside assistance.
She said her students appear to fall into two broad groups: one that continues to engage with the material and another that is losing critical-thinking skills. Bruckman described the second group as roughly the bottom half of her class, but the available information does not include a formal measurement, sample size, course, time period or independent review of that conclusion.
“The bottom half are definitely losing their critical thinking skills,” Bruckman said. “We’re not losing everyone, but we’re losing a good chunk.”
Her approach is one response among several being considered by educators. Some instructors are using handwritten, timed or supervised exams. Others favor open-ended projects, demonstrations, presentations and other performance-based work that asks students to show how they reached an answer.
These approaches are not part of one universal policy. They reflect a broader effort to make assessment measure learning rather than simply the quality of a final product.
What educators and students fear
Several surveys point to concern about dependence on AI, although the surveys cover different groups and should not be compared as if they measured the same population.
A College Board report found that 89% of surveyed principals worried AI could make students dependent on technology for basic tasks. Eighty-seven percent worried it could weaken the development of critical-thinking skills, and 82% worried it could interfere with deep engagement with course material.
A survey of 1,057 college faculty members by the American Association of Colleges and Universities and Elon University’s Imagining the Digital Future Center found that 90% believed AI would reduce students’ critical-thinking abilities. Ninety-five percent believed it would make students overly reliant on AI.
Students have expressed similar concerns. A RAND survey found that 70% of college students surveyed worried that AI use could harm their critical-thinking skills.
Other survey findings focus on the practical problem instructors face: measuring student learning. One survey cited by The Conversation found that 53% of national respondents were concerned about assessing learning when AI was used. Forty percent reported increased student reliance on AI, while 33% reported reduced critical thinking or problem-solving.
The survey methods and full questionnaires for some of these findings were not provided in the available material. The populations also differ: principals, faculty members and students may have different experiences and views of AI.
Ban or redesign?
Bruckman said banning AI at universities may not prepare students for workplaces where the technology is likely to be available. Instead, she supports redesigning courses and assessments so students learn when and how to use AI—and when they must work without it.
“I think we have to redesign our assessments to be resistant to AI,” she said.
That does not necessarily mean eliminating essays, projects or digital tools. It can mean adding steps that show a student’s reasoning, such as drafts, oral explanations, in-class writing or follow-up questions. It can also mean assessing skills that an AI system cannot demonstrate on a student’s behalf, including judgment, reflection and the ability to connect ideas to a specific problem.
The shift changes daily work for both teachers and students. Teachers must create assignments that reveal the learning process, set clear rules for permitted AI use and decide when direct observation is needed. Students may face more supervised work and more requests to explain their choices, not just submit a finished answer.
The available research does not identify a Washington, D.C., school, university, district policy or effective date tied to these changes. It does show why the debate is moving beyond academic dishonesty: educators are trying to determine whether technology is supporting learning or replacing the thinking that education is supposed to develop.
