This article covers SpringIQ: AI Detection: what it analyzes, how to enable it, and how to read results. For AI-based rubric scoring, see SpringIQ: Rubric (AI Scorecards).
Overview
SpringIQ: AI Detection is part of AwardSpring's SpringIQ intelligence suite. It analyzes student essays and short-answer responses in your General Application and Additional Requirements for indicators of AI-generated writing, providing administrators and reviewers with a confidence-based signal to inform their review process.
How It Works
When AI Detection is enabled and a student submits or updates an essay or short-answer response, AwardSpring automatically analyzes the text using industry-standard large language model (LLM) technology. The analysis looks for patterns typical of AI-generated writing, such as:
Overly polished or generic phrasing
Lack of personal detail or authentic voice
Sudden tonal or stylistic shifts
Unnatural fluency or repetition
Vague generalities without specific examples
The system returns a confidence level representing the likelihood that AI was involved in generating the response.
Important Distinctions
Confidence, not quantity. The confidence level represents how likely it is that AI was used in the response. It does not indicate how much of the text was AI-generated.
Informational, not deterministic. AI Detection does not automatically reject applications, reduce review scores, or take any action on your behalf. It provides a signal for your team to interpret during the review process.
Probabilistic, not certain. Treat results as one data point among many. Strong writers may occasionally receive elevated scores, and sophisticated AI usage may not always be detected.
Supported Question Types
AI Detection works on:
Essay Answer questions
Open-end Answer questions
AI Detection does not work on:
PDF or file upload responses
Multiple-choice, dropdown, or other non-text question types
Enabling AI Detection
AI Detection can be enabled for open-ended or essay-style questions in the following areas:
General Application
Before Applying Additional Requirements within a specific scholarship:
Steps to Enable
Navigate to the question you would like to update or create a new question.
Select one of the supported question formats for AI Detection (essay or short-answer).
Select the Enable for student responses checkbox.
Once enabled, AI Detection will automatically analyze all new essay and short-answer submissions moving forward.
Optional: Warn Applicants
You may also select the Warn applicants about AI detection checkbox. When selected, applicants see the message "AI Detection is enabled." above the response box, letting them know their responses will be analyzed for AI-generated content. This can help discourage the use of AI before submission.
Important: AI Detection is not retroactive. Responses submitted before the feature is enabled will not be analyzed. Only new or updated submissions after activation will be included.
Understanding Results
Confidence Levels
AI Detection results are displayed as one of four confidence levels:
Confidence Level | Threshold | What It Means |
Low | Below 33% | Low likelihood that AI was used in this response. |
Moderate | 34% – 66% | Some patterns suggest possible AI involvement. Consider reviewing more closely. |
High | 67% – 90% | Multiple indicators suggest likely AI involvement. Recommend careful review. |
Very High | Above 90% | Strong indicators of AI-generated content. Strongly recommend close examination. |
These thresholds are currently fixed across all accounts.
Where Results Appear
AI Detection results are visible in the following locations:
Student Application View (Admin perspective)—when viewing a student's application from Users > Application within a student profile, the AI Detection confidence level is displayed alongside each analyzed essay or short-answer response.
Award & Deny (A&D) Screen—AI Detection indicators appear with student responses
Reviewer View—reviewers assigned to review groups will see AI Detection results when reviewing applications that contain analyzed responses.
Note: AI Detection results are never visible to students. Only administrators and reviewers can see these signals.
Frequently Asked Questions
Can I customize the confidence level thresholds (e.g., change what counts as "High")?
Not at this time. The thresholds are globally fixed.
Can I hide AI Detection results from reviewers?
Not currently. When AI Detection is enabled, administrators and reviewers can see the results. We are evaluating visibility controls for a future update.
Does AI Detection run on answers that were submitted before I enabled the feature?
No. AI Detection analyzes only responses submitted or updated after you enable the feature.
Does AI Detection work on PDF/file uploads?
No. AI Detection analyzes only text entered directly into the application's essay and short-answer text boxes.
Will AI Detection automatically reject an application or lower a student's score?
No. AI Detection is purely informational. It provides a confidence-based signal but does not automatically take action on applications or scores.
Tips for Getting the Most Out of AI Detection
Use it as one signal among many. AI Detection is most effective when combined with your team's judgment, review group discussions, and other evaluation criteria.
Warn applicants. Even a warning can significantly reduce AI use in applications.
Educate your review team. Make sure reviewers understand that confidence levels indicate the likelihood of AI involvement, not certainty. A "High" result warrants closer examination, not automatic disqualification.
Consider your institution's AI policy. AI Detection works best when paired with a clear institutional policy on acceptable AI use in scholarship applications. Communicate that policy to students through your application instructions.