This article takes a look at how AI tools like ChatGPT have changed the way college grades work—especially in classes that rely on homework and other out-of-class assignments. It also digs into what this shift means for teaching, assessment, and academic honesty in higher education.
AI, grades, and learning patterns
Since 2022, when ChatGPT hit the mainstream, colleges have noticed a clear bump in excellent grades—about 30%—in courses where AI can actually help, like English composition and coding. But in more hands-on areas, such as sculpture or lab classes, there’s been no similar jump, which makes sense since those subjects rely more on in-person work than take-home tasks.
Igor Chirikov, a UC Berkeley professor, dug into grade data from 2018–2025 at a large, selective Texas research university (they didn’t say which one). He found that a lot of students moved from C-level work to A-level results when they used AI. He suspects this isn’t just a fluke at one school, but part of a bigger trend in higher ed.
The strongest effects from AI show up in classes with lots of homework, projects, and unsupervised assignments. Basically, students seem to get a boost on out-of-class work when they’re able to use AI as part of their process.
Where AI helps—and where it falls short
The numbers show that AI-enabled learning gives students a leg up on take-home work, probably because technology can stretch study time and give instant feedback. Still, this isn’t true for every subject. In hands-on fields that demand physical skills or studio practice, AI doesn’t seem to move the needle much, highlighting where digital tools just can’t replace real-world experience.
There’s a bigger question floating around here: do higher grades really mean students understand the material, or are they just getting better at using AI to turn in the “right” answers? Some faculty worry this could mean graduates aren’t actually ready to work independently—or that they’ll just lean on AI without really knowing their stuff.
Implications for students, instructors, and institutions
This trend of higher grades thanks to AI connects with old issues in higher ed, like grade inflation and how student feedback can influence promotions. Grade inflation isn’t new, but AI might be turning up the heat if tests and assignments don’t actually check for real understanding. There’s also a lot of talk among faculty about academic honesty, since unchecked AI use can make it hard to tell what’s original work and what’s not.
Practical assessment strategies in the AI era
- Design AI-integrated assignments where students can use LLMs, but they have to clearly document how they used them and show what they understand.
- Balance formats by mixing AI-enabled assignments with handwritten or oral parts to check both the process and the final product.
- Implement progressive assessment that tracks skill growth over time, using rubrics that value critical thinking and problem-solving more than just the end result.
- Don’t just punish misuse; instead, focus on creative evaluation strategies that show how students learn and grasp concepts.
- Offer faculty development so instructors can spot AI-generated work, build better assessments, and push back against grade inflation with thoughtful teaching.
Moving forward: policy and pedagogy
Experts like Chirikov say there’s no magic fix for AI-driven grade changes. Universities really need to try creative, evidence-based assessment strategies that let students use AI as a tool, but still demand real mastery of the subject. That means rethinking incentives, putting resources into teacher training, and coming up with assignments that make students explain their reasoning, sources, and exactly how AI played a part in their work.
Recommended actions for universities and faculty
- Adopt transparent AI-use policies that clearly state when and how people can use AI. These policies should also require everyone to disclose and explain any AI contributions.
- Diversify assessment modalities to capture a range of skills. Think critical thinking, design thinking, practical execution, and communication—don’t just stick to one method.
- Rethink incentives by focusing more on mastery-based outcomes in promotions and evaluations. Let’s not rely only on student satisfaction or standardized grades.
- Invest in ongoing pedagogy research to test and refine assessments. The goal is to find a balance between using AI and making sure students really learn something.
Here is the source article for this story: The ChatGPT era prompts a boom in A-graded coursework