MétaCan
Menu
← Back to cohort
Record W4401313066 · doi:10.18260/1-2--48322

WIP: Traditional Engineering Assessments Challenged by ChatGPT: An Evaluation of its Performance on a Fundamental Competencies Exam

2024· article· en· W4401313066 on OpenAlexaff
Trini Balart, Jorge A. Baier, Martín Castillo

Bibliographic record

Venuenot available
Typearticle
Languageen
FieldMedicine
TopicArtificial Intelligence in Healthcare and Education
Canadian institutionsUniversity of Toronto
Fundersnot available
KeywordsContext (archaeology)Coherence (philosophical gambling strategy)Relevance (law)Computer scienceInclusion (mineral)Engineering educationFunction (biology)Mathematics educationEngineering managementPsychologyEngineeringMathematicsStatistics

Abstract

fetched live from OpenAlex

Abstract ChatGPT, a chatbot which produces text with remarkable coherence, is leading higher education institutions to question the relevance of the current model of engineering education and, particularly, assessment. Among the many reasons behind this questioning is the fact that ChatGPT has been shown to be able to pass various engineering exams. In this research, the GPT-3.5 and GPT-4 models were used to solve different real-life versions of the Fundamental Competencies Exam (FCE), an exam used by a selective Latin American engineering school upon the completion of quintessential engineering courses like basic dynamics, ethics for engineers, and probability and statistics. The formulation of the questions seeks to demonstrate that the student has the fundamental knowledge of the discipline. We adopted a strategy in which the questions were extracted from the FCE modules, and translated to LaTeX. The statements of each question were presented in the absence of supplementary context to avoid influence between questions within the same exam. In addition, a comparative analysis of the effectiveness of the GPT-4 model is performed, evaluating its performance with and without image interpretation capability, due to the recent inclusion of the multimodal function of ChatGPT-4. The results obtained reveal that the difference in the pass rate of GPT-3.5 and GPT-4 is considerable, with 47.38% and 63.06% respectively. While the GPT-4 version without images achieved a passing rate sufficient to pass the exam in all modules, the results that include questions with images increased even more, reaching a 64.38% pass rate. We must continue to solve different versions of the exam. These data will allow us to perform multiple analyses related to the historical performance of the exam, providing a proxy to assess how difficulty has changed over the years. In light of these preliminary results, and given the tight constraints imposed on the model, it is imperative to question whether the FCE effectively assesses the fundamental skills required for an engineer and whether it is the best method for assessing foundational engineering competencies amidst the advent of innovative AI tools.

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 imitation

Not 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.

metaresearch head score (Codex)0.005
metaresearch head score (Gemma)0.024
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Bench or experimental · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.005
Threshold uncertainty score0.025

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0050.024
Meta-epidemiology (narrow)0.0010.000
Meta-epidemiology (broad)0.0010.001
Bibliometrics0.0020.001
Science and technology studies0.0000.000
Scholarly communication0.0010.001
Open science0.0010.002
Research integrity0.0010.001
Insufficient payload (model declined to judge)0.0040.003

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.

Opus teacher head0.424
GPT teacher head0.463
Teacher spread0.040 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designBench or experimental
Domainnot available
GenreEmpirical

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".

Quick stats

Citations0
Published2024
Admission routes1
Has abstractyes

Explore more

Same topicArtificial Intelligence in Healthcare and Education→French-language works237,207→