Performance of AI Chatbots on the fundamentals of engineering civil exam
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
The growing integration of artificial intelligence (AI) in education, particularly through AI chatbots powered by large language models (LLMs), requires careful evaluation of their benefits and limitations. This study examines the potential of three leading AI chatbots—ChatGPT, Gemini, and DeepSeek—as educational tools for civil engineering students by evaluating their performance on the fundamentals of engineering (FE) civil exam. Using a standardized dataset, chatbot responses were analyzed across three criteria: Final Answer Correctness, Conceptual Understanding, and Correct Use of Equations. Indicative results show that ChatGPT o3, ChatGPT-4o, DeepSeek-R1, and Gemini 2.5 Pro achieved accuracies above 70%, while DeepSeek-V3 and Gemini 2.0 Flash scored above 60%. Performance was highest in foundational subjects introduced early in engineering curricula and lowest in advanced, domain-specific areas, indicating the need for enhanced reasoning capabilities and targeted domain training. In addition, the performance of AI chatbots was further analyzed by comparing their accuracy on text-based versus image-based questions. Accuracy was significantly higher for text-based questions (average 87%) compared to image-based questions (42%), revealing current limitations in visual interpretation. These findings suggest that while AI chatbots can potentially serve as practical tutoring tools for early-stage learners, further refinement is needed for complex, visual, or advanced engineering tasks. This study contributes to understanding the role of AI in civil engineering education and informs strategies for its integration into academic practice.
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 | 0.001 |
| 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