Related Experiment Video
Updated: Jan 18, 2026
A Postoperative Evaluation Guideline for Computer-Assisted Reconstruction of the Mandible
Published on: January 28, 2020
AI-assisted translation in oral and maxillofacial surgery: A comparative evaluation
Murillo Martins Pinto1, Erick Elienai Reis Araújo1, Zacharie Rahhal2
1School of Dentistry, University of São Paulo, São Paulo, Brazil.
Generative artificial intelligence (AI) translation tools show promise for non-native English speakers in scientific publishing. DeepSeek-V3 excelled in overall performance, clarity, and delivery, offering a valuable aid for researchers.
Area of Science:
- Computational Linguistics
- Scientific Communication
- Artificial Intelligence Applications
Background:
- English is the predominant language in scientific research, creating barriers for non-native English speakers.
- Generative artificial intelligence (AI) translation tools present a potential solution to improve accessibility and publication rates.
- Systematic evaluation of AI translation tool performance is crucial for assessing their suitability in academic contexts.
Purpose of the Study:
- To evaluate the performance of leading generative AI translation tools.
- To determine the suitability of these AI tools for non-native English-speaking researchers aiming to publish scientific work.
- To compare AI translation quality across different genres (technical, academic, descriptive).
Main Methods:
- Thirty non-English texts were translated using six AI tools: ChatGPT-4.0, Claude 3.7, Copilot, Gemini 2.0, DeepSeek-V3, and Perplexity.
- Translations were assessed for correctness, clarity, engagement, and delivery using Grammarly®.
- Statistical analyses (ANOVA, Kruskal-Wallis) were employed to compare the performance of the AI tools.
Main Results:
- DeepSeek-V3 achieved the highest overall score (mean=92.9), significantly outperforming Claude and Copilot, and matching Gemini.
- Correctness varied significantly across tools, with Copilot showing strong performance, while DeepSeek-V3 led in clarity, engagement, and delivery.
- Gemini 2.0 ranked second but demonstrated limitations in iterative refinement, often deeming its initial translations as final.
Conclusions:
- DeepSeek-V3 emerged as the most effective AI translation tool for scientific texts, with Gemini 2.0 as a close second.
- AI translation tools can significantly reduce communication barriers for non-native English-speaking researchers.
- Continued transparency and refinement of AI translation technologies are necessary to balance efficiency with academic integrity.
More Related Videos
07:32Author Spotlight: 3D Movement Assessment of Maxillary Posterior Teeth in Clear Aligner Treatment
Published on: February 23, 2024
05:54Real-Time Dynamic Navigation System for the Precise Quad-Zygomatic Implant Placement in a Patient with a Severely Atrophic Maxilla
Published on: October 18, 2021
