Skip to main content Skip to navigation

Steering AI-Driven Personalization of Scientific Text for General Audiences

Project Overview

The document examines the role of generative AI in education, particularly through the development of TranSlider, an innovative AI-powered tool aimed at personalizing scientific texts for broader audiences. It addresses the prevalent challenges in science communication, particularly the disparities in scientific literacy among different audience groups. TranSlider empowers users to customize the degree of personalization in translations, thereby improving comprehension and engagement with complex scientific material. User studies revealed that participants found the tool's personalized translations and interactive features valuable, enhancing their learning experience. However, the findings also raised concerns regarding the reliability of AI-generated content and the potential for diminished access to original source materials, highlighting the need for careful consideration in the deployment of AI tools in educational contexts. Overall, the document underscores the potential of generative AI to facilitate better understanding and accessibility of scientific information while acknowledging the challenges that accompany its use in educational settings.

Key Applications

TranSlider

Context: General audiences seeking to understand scientific texts

Implementation: An interactive slider allows users to adjust the degree of personalization in translations of scientific articles based on user profiles.

Outcomes: Participants reported improved understanding of scientific content and appreciated the relatable and contextual translations.

Challenges: Concerns about the reliability of AI-generated content and the potential for users to depend on simplified translations instead of engaging with original materials.

Implementation Barriers

Reliability

Concerns regarding the consistency and trustworthiness of AI-generated personalized translations.

Proposed Solutions: Encouraging readers to verify translations against original sources and involving authors in verifying AI-generated translations.

Access to Original Material

The risk that personalized translations could discourage users from accessing original scientific texts.

Proposed Solutions: Balancing the use of AI translations with strategies to maintain interest in original documents, emphasizing the importance of deeper understanding.

Project Team

Taewook Kim

Researcher

Dhruv Agarwal

Researcher

Jordan Ackerman

Researcher

Manaswi Saha

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Taewook Kim, Dhruv Agarwal, Jordan Ackerman, Manaswi Saha

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies