Chatting in Academia? The impact of large language models and conversational agents on literary studies

Proposal

Since the launch, in November 2022, of the famous conversational agent built by OpenAI, both the general public and politicians have discovered the existence of large language models (LLM) and deep learning. More or less fantasized projections are made about this type of tool, displaying both unprecedented capacities (especially in terms of coherence and even elegance in writing), and incomprehensible failures ("hallucinations" like the one in subject of the sheep eggs).

The question that arises is that of the real contribution that such tools can have in different sectors. The conference organized by the ARIANE Huma-Num french consortium wishes to focus on the perspectives opened up by major language models, coupled or not with conversational agents, for literary studies. We will understand the latter in a very broad sense, relating both to texts recognized for their aesthetic value, whether or not they are inscribed in a canon, and to texts that do not have such ambition, mobilized in historical, philosophical, linguistic studies, etc.

We invite you to submit a communication proposal if you use or have used an LLM and/or a conversational agent for:

- write a summary, a synthesis, any type of text related to your scientific work;

- annotate texts, whatever the output format of the observables (xml, json, spreadsheet, etc.)

- generate or correct computer code to carry out your work;

- carry out a study of the literature, or a state of the art;

- analyze your data, thanks to an additional machine learning step (fine-tuning);

 etc.

Communications should focus on the interest and limits of the use of LLMs/conversational agents, as well as on the different stages, or the various protocols, which had to be put in place to obtain results. Presentations of possible failed experiments are just as welcome, especially coupled with the presentation of the alternatives adopted: manual work, use of older software, traditional machine learning, etc.

Three types of proposals are expected:

  1. Communications lasting around twenty minutes, followed by questions. Proposals of 1500 words (excluding bibliography), but including a brief presentation of the author(s).
  2. Round tables lasting 1h30 to 2h. Round tables bring together between 4 and 8 participants (including the moderator). Proposals of 3000 words (excluding bibliography), but including a brief biography of each speaker.
  3. Posters. Proposals of 500 words (excluding bibliography), including a brief presentation of the main author.

Communications must be submitted before September 25, 2024 (6 p.m.).
The acceptance (or reject) will be notified by September 30 at the latest.
The conference will take place on November 27 and 28, 2024, at the Campus Condorcet, Aubervilliers.

Scientific committee

Ioana Galleron
Fatiha Idmhand
Sabine Loudcher
Nathalie Rousseau
Stéphanie Dord-Crouslé
Alexei Lavrentiev
Motasem Alrahabi
Anne-Emmanuelle Kahn

Organization committee

Ioana Galleron
Fatiha Idmhand
Sabine Loudcher
Anaïs Chambat
Mathilde Le Meur
Andrès Felipe Echavarria



Online user: 2 Privacy
Loading...