Aarhus University Seal

Events

CLAI talks: Ulf Dalvad Berthelsen and Ea Lindhardt Overgaard present the joint project: “Do Large Language Models Have a Disciplinary Voice?", May 16

On May 16, Ulf Dalvad Berthelsen and Ea Lindhardt Overgaard, Aarhus University, will give a CLAI-talk on the project entitled: "Do Large Language Models Have a Disciplinary Voice?"

Info about event

Time

Thursday 16 May 2024,  at 13:30 - 14:00

Location

Virtual

Organizer

Center for Language Generation and AI

On May 16, Ulf Dalvad Berthelsen and Ea Lindhardt Overgaard, Aarhus University, will give a CLAI-talk on their joint project entitled: "Do Large Language Models Have a Disciplinary Voice?"

A comparative corpus-based study of GPT4’s ability to reproduce disciplinary voice in AI-generated linguistic prose.

The purpose of our study is to uncover the extent to which generative AI models - with GPT4 as an example - can reproduce disciplinary voice in academic prose written in Danish. As these models are trained on vast amounts of natural language, we must, from a functional linguistic point of view, expect that the genre and register variation we find in natural language is reproduced to some extent in the auto-generated output. We are particularly interested in the phenomenon of disciplinary voice, partly because it is assumed to be a difficult feature to reproduce, and partly because it is a relatively well-described phenomenon that can be investigated quantitatively through analysis of the surface structure of texts. We focus specifically on three aspects of disciplinary voice: stance, engagement, and subject-specific vocabulary. We investigate the phenomenon quantitatively through a corpus-based comparative study in which we compare a corpus consisting of linguistic articles written in Danish with a corpus of AI-generated academic prose with linguistic content.

 


The Center for Language Generation and AI at Aarhus University is committed to fostering the interchange of ideas and support for researchers in the area of language generation and AI. By opening our biweekly talks to the public, we aim to create a platform for knowledge sharing and collaboration among researchers, students, and the broader community.

These talks will be conducted virtually, allowing attendees from around the world to participate. Registration details and links to join each talk will be provided on our website. All talk will take place at 13:30 (CEST) on Zoom. The link ishttps://aarhusuniversity.zoom.us/j/66002478758

For more information and updates on our biweekly talks, please check our news page or follow us on LinkedIn or Twitter.

For media inquiries and questions, please contact: pascale.moreira@cc.au.dk

 

Past events

CLAI talks: Ulf Dalvad Berthelsen and Ea Lindhardt Overgaard present the joint project: “Do Large Language Models Have a Disciplinary Voice?", May 16

On May 16, Ulf Dalvad Berthelsen and Ea Lindhardt Overgaard, Aarhus University, will give a CLAI-talk on the project entitled: "Do Large Language Models Have a Disciplinary Voice?"

Info about event

Time

Thursday 16 May 2024,  at 13:30 - 14:00

Location

Virtual

Organizer

Center for Language Generation and AI

On May 16, Ulf Dalvad Berthelsen and Ea Lindhardt Overgaard, Aarhus University, will give a CLAI-talk on their joint project entitled: "Do Large Language Models Have a Disciplinary Voice?"

A comparative corpus-based study of GPT4’s ability to reproduce disciplinary voice in AI-generated linguistic prose.

The purpose of our study is to uncover the extent to which generative AI models - with GPT4 as an example - can reproduce disciplinary voice in academic prose written in Danish. As these models are trained on vast amounts of natural language, we must, from a functional linguistic point of view, expect that the genre and register variation we find in natural language is reproduced to some extent in the auto-generated output. We are particularly interested in the phenomenon of disciplinary voice, partly because it is assumed to be a difficult feature to reproduce, and partly because it is a relatively well-described phenomenon that can be investigated quantitatively through analysis of the surface structure of texts. We focus specifically on three aspects of disciplinary voice: stance, engagement, and subject-specific vocabulary. We investigate the phenomenon quantitatively through a corpus-based comparative study in which we compare a corpus consisting of linguistic articles written in Danish with a corpus of AI-generated academic prose with linguistic content.

 


The Center for Language Generation and AI at Aarhus University is committed to fostering the interchange of ideas and support for researchers in the area of language generation and AI. By opening our biweekly talks to the public, we aim to create a platform for knowledge sharing and collaboration among researchers, students, and the broader community.

These talks will be conducted virtually, allowing attendees from around the world to participate. Registration details and links to join each talk will be provided on our website. All talk will take place at 13:30 (CEST) on Zoom. The link ishttps://aarhusuniversity.zoom.us/j/66002478758

For more information and updates on our biweekly talks, please check our news page or follow us on LinkedIn or Twitter.

For media inquiries and questions, please contact: pascale.moreira@cc.au.dk