Using GPT-4 Turbo to Automatically Identify Defeaters in Assurance Cases
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Assurance cases (ACs) are convincing arguments, supported by a body of evidence and aiming at demonstrating that a system will function as intended. Producers of systems can rely on assurance cases to demonstrate to regulatory authorities how they have complied with existing industrial standards (e.g., ISO 26262, DO-178C). Defeaters are arguments that challenge the effectiveness of assurance cases. Their presence in assurance cases could compromise the reliability of these assurance cases and make them inadequate for verifying a system's capabilities (e.g., safety, and security). This may lead to system failure, which could have severe outcomes, including loss of life. Therefore, identifying and mitigating defeaters is key to improving assurance cases robustness and reliability. In this paper, we focus on the identification of defeaters. Thus, we rely on GPT-4 Turbo, a Large Language Model developed by OpenAI, to automate the generation (identification) of defeaters in assurance cases. Our approach uses the Eliminative Argumentation (EA) notation to represent assurance cases. Besides, we leverage the Chain of Thought prompting technique to improve GPT-4 Turbo's reasoning capabilities. We conducted experiments on various reference assurance case fragments from the nuclear and aviation domains to evaluate the ability of GPT-4 Turbo to automatically generate defeaters. Although the quality of our experiments results is relatively moderate, the analysis of these results still provides valuable insights on the effectiveness of GPT-4 Turbo in generating defeaters.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 it