In today’s classrooms, artificial intelligence is no longer a future concept; it’s a present reality. Recent questions about how students are using AI in their classes have included questions focusing on how much of students’ work is their own and how much is generated by AI. Whether through drafting lab conclusions, generating practice quizzes, or organizing essay ideas, AI is now embedded in how students approach learning. AI’s growing presence raises a pivotal question: How do we ensure that assessment still reflects what students know, understand, and can do?

Traditionally, science educators have looked for evidence of student understanding through products such as worksheets, essays, or lab reports. However, in the AI era, the question is no longer just about whether students can produce answers. Now we must ask these questions: Did the student truly create this? Did they engage in the thinking, the reasoning, and the reflection that drive meaningful learning?

This shift requires science educators to move away from a product-centered approach and prioritize the learning process instead. In doing so, we begin to capture something far more valuable than a correct answer; we begin to capture students’ growth. 

Understanding the Shift: What the Research Tells Us

Educational research has long emphasized the need for assessments that go beyond rote memorization. Scholars such as Holmes, Bialik, and Fadel (2019) argue that in a world shaped by AI, assessments must evolve to prioritize creativity, critical thinking, and collaboration—all skills that AI can assist with, but cannot replicate. Likewise, Merrill’s (2002) Principles of Instruction underscore that learning deepens when students activate prior knowledge, apply concepts to real-world problems, and integrate those experiences into lasting understanding.

In science education, this shift is not new and has already been underway. The Next Generation Science Standards (NGSS) emphasize the importance of scientific practices, asking students to explain phenomena, use models, and engage in argumentation based on evidence. These standards encourage educators to go beyond the “what” and instead focus on the “how” and “why,” questions that become especially relevant in the presence of AI.

The Problem With Traditional Assessment in an AI World

Many classic assessments—multiple-choice tests, fill-in-the-blank worksheets, fact-based prompts, and regurgitation of information—can now be completed with minimal effort using AI tools. A chatbot can label a diagram or define Newton’s laws. It can even draft a decently formatted lab report. But it can’t reflect on a failed experiment or explain a moment of insight during data analysis. 

In this new landscape, we need to design assessments that are both harder to outsource to technology and more engaging for students. That means asking them to build, create, test, revise, and defend their thinking. It means rewarding not just the end product, but the journey that led to it.

Take, for example, a common biology assignment. Instead of asking students to list the parts of a cell, we might ask them to create a cartoon format outline in which a story is presented from the perspective of a cell organelle, explaining how it supports life within the larger system. Or rather than grading a physics quiz on Newton’s laws, we could challenge students to design a playground slide and explain how those laws influence safety features. These types of tasks reveal whether a student understands the content deeply enough to apply it meaningfully, while also providing opportunities for creativity.

The challenge here is that students need to demonstrate their own work, in their own writing and illustration, that reflects their change in thinking over time.  So documentation of the progression of their understanding is far more important than a single file or product they produce at the end of a unit.

AI as a Formative Partner—Not a Shortcut

While some educators may feel threatened by AI, there’s another perspective worth considering: AI can actually strengthen formative assessment. When used responsibly, AI tools can provide students with immediate feedback, generate practice questions tailored to learning goals, or support multilingual learners drafting scientific explanations.

The difference lies in how the tools are framed. If students see AI as a ghostwriter, something to think for them, they miss out on growth. But if they treat AI as a thinking partner, one that challenges them, highlights errors, or proposes alternatives, then the learning remains front and center.

Imagine a student drafting a lab report and asking an AI tool to critique the clarity of their claim. The student can choose to accept, reject, or refine that feedback. If approaching AI from this perspective, it would be important for students to submit their original work, the prompt that they wrote for AI, the revised work using AI, and their reaction to the recommendations. For example, they might include a reflection like this one in their final submission: “The AI suggested removing this sentence, but I kept it, since it connects to the data in Figure 2 because...” That kind of metacognition is exactly what we want students to develop.

To support this strategy, educators can begin including short “Tool Use Declarations” in assignments. These are simple but insightful reflections that ask, “What tools did you use? What suggestions did you accept, and what did you ignore?” Not only does this encourage responsible AI use, but it also builds transparency and student ownership.

Designing Assessments That Reflect Real Thinking

Authentic assessment is not new in science education. Many teachers already ask students to design experiments, present findings, or reflect on lab results. However, in the AI era, these practices have become even more crucial.

Performance-based assessments, such as building prototypes, defending scientific claims orally, or curating digital portfolios, encourage students to demonstrate understanding in multi-dimensional ways. These formats naturally resist automation efforts that are part of AI use, as they rely on creativity, context, and the articulation of thought processes. A chatbot might generate a script, but it cannot present a model with spontaneous commentary, respond to a follow-up question, or describe the emotional experience of discovery.

Digital portfolios, in particular, offer exciting possibilities. A student could document their learning over several weeks, annotating entries with moments of insight or confusion. They might generate graphs or summaries using AI tools, but always with acknowledgment, noting how these tools helped (or hindered) their work. When assessment becomes a story of learning rather than a snapshot of recall, both teachers and students benefit.

Rethinking the Rubric: Grading for Growth

Our tools for measuring learning must also evolve. Traditional rubrics often emphasize specific requirements, a set number of examples, or at times, surface-level features: grammar, spelling, formatting. But in a world where AI excels at these tasks, we need to look deeper.

New rubrics should highlight reasoning, transfer, voice, and iteration. Did the student connect evidence to their claim? Did they apply their learning in a novel context? Is there evidence that they revised or rethought their ideas?

Merrill’s framework provides a useful structure here: Effective learning moves from activation to demonstration to application and finally to integration. Rubrics that follow this sequence help ensure that students are not just completing assignments, but also growing as thinkers.

Supporting Teachers Through the Shift

None of these changes can happen in isolation. Teachers need professional learning opportunities that go beyond tool tutorials. We need space to reflect, experiment, and collaborate on new assessment strategies. We need time to revise rubrics, test new formats, and talk with colleagues about what works—and what doesn’t.

We also need support from administrators and policy leaders. Schools must develop clear, transparent guidelines about AI use, helping students and families understand what’s allowed and what’s not, and why. Any systems that auto-score writing or open responses must be transparent in how they work, and teachers must be empowered to challenge errors or biases.

Moving Forward: Assessment With Integrity and Innovation

Every wave of educational technology reshapes what learning looks like. Chalkboards, calculators, and online learning have all changed how we teach and assess. Generative AI may be the most disruptive shift yet, but it’s also an opportunity.

By rethinking assessment now, we can help students become not just content experts, but also agile, ethical, and reflective thinkers. We can create systems that value process over presentation, and growth over grades. We can harness AI not as a threat, but as a tool that when used wisely, enhances both teaching and learning.

In this new frontier, there are no perfect answers. But there are essential questions: How do we keep learning authentic? How do we build systems that value curiosity, reflection, and human agency?

These questions and others are the ones that we as educators need to reflect on in this age of AI assessment.

References

Holmes, W., M. Bialikand C. Fadel. 2019. Artificial Intelligence in Education: Promises and Implications for Teaching and Learning. Boston, MA: Center for Curriculum Redesign.

Merrill, M. D. 2002. First principles of instruction. Educational Technology Research and Development 50 (3): 43–59.

Christine Anne Royce, Ed.D., is a past president of the National Science Teaching Association and currently serves as a Professor in Teacher Education and the Co-Director for the MAT in STEM Education at Shippensburg University. Her areas of interest and research include utilizing digital technologies and tools within the classroom, global education, and the integration of children's literature into the science classroom. She is an author of more than 140 publications, including the Science and Children Teaching Through Trade Books column.

Valerie Bennett, Ph.D., Ed.D., is an Assistant Professor in STEM Education at Clark Atlanta University, where she also serves as the Program Director for Graduate Teacher Education and the Director for Educational Technology and Innovation. With more than 25 years of experience and degrees in engineering from Vanderbilt University and Georgia Tech, she focuses on STEM equity for underserved groups. Her research includes AI interventions in STEM education, and she currently co-leads the Noyce NSF grant, works with the AUC Data Science Initiative, and collaborates with Google to address CS workforce diversity and engagement in the Atlanta University Center K–12 community.

Note: This article is part of the blog series From Chalkboards to AI, which focuses on how artificial intelligence can be utilized within the classroom in support of science as explained and described in A Framework for K–12 Science Education and the Next Generation Science Standards. 

The mission of NSTA is to transform science education to benefit all through professional learning, partnerships, and advocacy.