Using AI with Machine Learning Models to Predict Concrete Compressive Strength

10 pages•Published: June 2, 2026

Tamer Breakah, Mohsen Goodarzi, Mohamed Abdel Raouf and Tarek Mahfouz

Abstract

This paper evaluates the performance of AI generated machine learning models. Three commonly used AI platforms were selected for this study; ChatGPT, Gemini, and Perplexity. The dataset used was uploaded to the AI platforms and the choice of models was left open. The number of models investigated by Perplexity was the highest. Another iteration of the modelling was done by asking the AI platforms to run the same analysis using nine different models and the models were specified in this case. The results showed different prediction accuracy among the models, while the prediction accuracy was close among the platforms, in most cases. The drop in prediction accuracy from the training set to the test set was also checked for all the models and platform to study the overfitting tendency. The results showed that Perplexity had the highest overfitting tendency, while ChatGPT had the lowest. The results show that the use of AI worked well with the different machine learning techniques. On the other hand, the regression models did not work well and require human intervention to make them work better.

Keyphrases: artificial intelligence, concrete compressive strength prediction, machine learning

In: Wesley Collins, Anthony Perrenoud and John Posillico (editors). Proceedings of Associated Schools of Construction 62nd Annual International Conference, vol 7, pages 1112-1121.

Links:	https://easychair.org/publications/paper/RSdV
	https://doi.org/10.29007/dj14

BibTeX entry

@inproceedings{ASC2026:Using_AI_with_Machine,
  author    = {Tamer Breakah and Mohsen Goodarzi and Mohamed Abdel Raouf and Tarek Mahfouz},
  title     = {Using AI with Machine Learning Models to Predict Concrete Compressive Strength},
  booktitle = {Proceedings of Associated Schools of Construction 62nd Annual International Conference},
  editor    = {Wesley Collins and Anthony Perrenoud and John Posillico},
  series    = {EPiC Series in Built Environment},
  volume    = {7},
  publisher = {EasyChair},
  bibsource = {EasyChair, https://easychair.org},
  issn      = {2632-881X},
  url       = {/publications/paper/RSdV},
  doi       = {10.29007/dj14},
  pages     = {1112-1121},
  year      = {2026}}

Download PDF Open PDF in browser