Prompt-Engineering Strategies for Minimizing Bias in Large Language Model Outputs: Applications in Computing Education
Bibliographic record
Abstract
As large language models (LLMs) increasingly permeate educational applications, concerns about the perpetuation of bias persist. We present our preliminary work on developing prompt-engineering strategies to mitigate bias in content generated by LLMs in computer science (CS) education. This work investigates both empirical insights into fairness-aware prompt formulation and actionable takeaways for educators. We focus on an initial list of prompting strategies for mitigating bias and explore their impact on educational content generation. Recent research has shown the efficacy of prompt-base debiasing [1] as well as the potential disadvantages of using prompts that have not been mitigated for bias, from user dissatisfaction [2] to unsafe outputs [5, 6]. Additionally, a growing body of empirical work points to the idea that certain properties of in-context examples such as flow [7], illustration [3], and order [4] could either improve or derail LLM performance. Our study leverages these findings in the context of generating educational content. The goal is to promote fairness-aware approaches which can be applied to the automated generation of learning materials and the development of LLM-based educational tools. This work also contributes practical insights on prompt-engineering to the evolving curriculum of Ethics in Artificial Intelligence (AI).
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.019 | 0.126 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.003 | 0.005 |
| Open science | 0.002 | 0.004 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.007 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".