Optimizing Natural Language Processing: A Comparative Analysis of GPT-3.5, GPT-4, and GPT-4o

Manuel Ayala-Chauvin and Fátima Avilés-Castillo

Data and Metadata, 2024, vol. 3, .359

Abstract: In the last decade, the advancement of artificial intelligence has transformed multiple sectors, with natural language processing standing out as one of the most dynamic and promising areas. This study focused on comparing the GPT-3.5, GPT-4 and GPT-4o language models, evaluating their efficiency and performance in Natural Language Processing tasks such as text generation, machine translation and sentiment analysis. Using a controlled experimental design, the response speed and quality of the outputs generated by each model were measured. The results showed that GPT-4o significantly outperforms GPT-4 in terms of speed, completing tasks 25% faster in text generation and 20% faster in translation. In sentiment analysis, GPT-4o was 30% faster than GPT-4. Additionally, analysis of response quality, assessed using human reviews, showed that while GPT-3.5 delivers fast and consistent responses, GPT-4 and GPT-4o produce higher quality and more de-tailed content. The findings suggest that GPT-4o is ideal for applications that require speed and consistency, while GPT-4, although slower, might be preferred in contexts where text accuracy and quality are important. This study highlights the need to balance efficiency and quality in the selection of language models and suggests implementing additional automatic evaluations in future research to complement the current findings

Date: 2024
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:dbk:datame:v:3:y:2024:i::p:.359:id:1056294dm2024359

DOI: 10.56294/dm2024.359

Access Statistics for this article

More articles in Data and Metadata from AG Editor
Bibliographic data for series maintained by Javier Gonzalez-Argote ().