Best Of

Academic integrity in the context of AI presents genuine challenges for instructors. From the start, it is essential to recognize that assignment design alone cannot fully eliminate student use of AI-generated text.

What instructors can do is design learning experiences where misrepresentation becomes difficult to sustain over time. Supporting integrity does not require monitoring every step of a student’s process or relying on unreliable detection tools. Instead, it requires making student thinking and decision-making visible in ways that AI-generated work struggles to align with consistently.

This article introduces four high-impact integrity checkpoints that can be implemented directly in Brightspace using tools instructors already know. Together, they provide credible authorship evidence, normalize ethical AI use, and reduce the likelihood of misunderstandings.

Design Goal: Minimum Effort, Maximum Clarity

These checkpoints are based on a simple principle:

If the learning process is visible, integrity becomes the default.

Rather than collecting exhaustive documentation, this approach captures just enough evidence at key moments to demonstrate how students are thinking, researching, and writing.

Checkpoint 1: Topic and Research Question Rationale

Purpose
This early checkpoint captures original intent and prevents integrity issues before they begin.

Tool choice note: For this checkpoint, an Assignment with a text submission is a practical choice. Assignments support exploratory writing, are faster to skim at scale, and align well with the low-stakes, process-oriented purpose of this checkpoint.

This early checkpoint captures original intent and prevents integrity issues before they begin.

What students submit

• A working topic
• A research question
• A short rationale (150–200 words) explaining why the question fits the assignment

AI use documentation
Students include one simple disclosure statement at the end of their submission:

• AI used for brainstorming or wording: Yes/No. If yes, briefly describe how.

Why this matters
Early thinking is difficult to fabricate convincingly with AI. Capturing it provides a clear baseline for authorship and helps instructors guide scope and direction early.

Checkpoint 2: Research Decision-Making

Purpose
This checkpoint makes research judgment visible, not just source collection.

Tool choice note: An Assignment works well when you want a private, assessable record of research decisions. A Discussion can be effective when you want to normalize research struggles and decision-making through peer visibility. Choose based on whether your goal is assessment or shared learning.

This checkpoint makes research judgment visible, not just source collection.

What students submit

• An annotated bibliography (3–5 sources) or
• A short research log describing key decisions

Required focus
Students address questions such as:

• Why was this source chosen or rejected?
• Identify one specific source that changed or refined your thinking. What did you originally think, and what changed after engaging with this source?
• Describe one topic change you made after starting your research (for example, narrowing the focus, shifting the angle, or dropping a subtopic) and explain what prompted that change. 
• Describe one question or line of inquiry you initially planned to pursue but later abandoned, and explain what made it less useful or relevant.

AI use documentation
Students explain where AI assisted, if at all, and how they verified or revised its output.

Why this matters
Evaluating sources and explaining decisions requires contextual reasoning that AI tools struggle to replicate authentically.

Checkpoint 3: Drafting Evidence Through Version History

Purpose
This is a passive safeguard that shows writing development over time.

Tool choice note: An Assignment with a file or link submission is the simplest way to collect drafting evidence without increasing workload. It allows instructors to access version history only if questions arise, rather than reviewing drafts routinely.

This is a passive safeguard that shows writing development over time.

What students submit

• A link to a Google Doc or Word Online draft with version history enabled

Student expectations

• Drafting must occur in a tool that records revision history
• Early, incomplete, and revised drafts are encouraged

AI use documentation
No separate form is required. Students may flag sections where AI assisted with clarity or editing.

Why this matters
Version history provides powerful evidence of authorship without adding to instructor workload. It only needs to be consulted if questions arise.

Checkpoint 4: Final Submission with Aligned Process Commentary

Purpose
This checkpoint consolidates integrity evidence by checking for alignment between the final paper and earlier process artifacts. It is not intended to serve as a standalone proof of authorship.

Tool choice note: An Assignment with both file and text submission fields allows students to submit the final paper and aligned commentary together, keeping integrity evidence consolidated in one place.

This checkpoint consolidates integrity evidence by checking for alignment between the final paper and earlier process artifacts. It is not intended to serve as a standalone proof of authorship.

What students submit

• Final paper
• A short aligned commentary (150–250 words)

Aligned commentary prompts
Students respond to prompts that explicitly reference earlier work:

• Identify one substantive change you made between an earlier draft and the final version and indicate where that change appears in the paper.
• Identify one source from your annotated bibliography or research log that most influenced your final claim and explain how that influence is visible.
• Identify one section of the paper that required the most reworking and briefly explain what earlier decision made it challenging.

AI use documentation
Students include a concise disclosure summarizing any AI use across the assignment stages.

Why this matters
Alignment-based commentary is harder to fabricate convincingly than generic reflections. Its value lies in how well it corresponds to documented drafts and research decisions, not in the commentary itself.

What This Approach Does and Does Not Prevent

This approach does not:

• Prevent students from using AI tools to generate text
• Guarantee that every submitted sentence is human-written
• Replace institutional academic integrity policies
• Eliminate the need for instructor judgment

This approach does:

• Make misrepresentation difficult to maintain across time and artifacts
• Surface inconsistencies between drafts, research decisions, and final work
• Normalize transparent, ethical AI use rather than concealment
• Reduce reliance on AI detection tools
• Provide defensible context if integrity questions arise

The goal is not AI-proofing. The goal is alignment-proofing: designing assignments that yield consistent traces of authentic learning and make shortcuts easier to detect through misalignment than through surveillance.

Final Takeaway

Supporting academic integrity in an AI-enabled world does not require more policing. It requires better visibility into learning.

By capturing these four moments in the writing process, instructors gain meaningful insight into student thinking, not just confirmation that a task was completed. These checkpoints are designed to support instructional judgment through visibility, not to function as compliance checks.

In doing so, instructors can protect both themselves and their students while reinforcing the core goal of higher education: meaningful thinking and learning.