International conference

9th Digital Document & Society Conference

INFORMATION and AI: OPPORTUNITIES and RISKS

September 12 and 13, 2024

at the Haute Ecole de Gestion (Geneva), co-organized by Dicen-IDF (Cnam) and HES-SO (HEG, Geneva)

  Version pdf

Scientific coordinators

Evelyne Broudoux, CNAM, Dicen-IDF

Ghislaine Chartron, CNAM, Dicen-IDF

Benoît Epron, HEG Geneva, HES-SO

Call for papers

Today, the term AI is used to describe the automation of computational processes, which originated in the design of neural networks (Lelu & Rosenblatt, 1986) and has reached a new level with deep learning models based on large language models (LLMs) (LINC-CNIL, 2023). Indeed, generative AI based on these models has entered the communicative sphere, with language interactions operated by conversational robots, capable of answering the most precise questions. This is leading to upheavals in access to information, which is itself changing nature, potentially created by algorithms. Authoriality and its associated rights are therefore being called into question (Broudoux, 2022).

Let's remember that these models can only work in "massive data" mode, the issues of which we had already explored in a previous edition of the Document et Société conference (Chartron & Broudoux, 2015). Based on deep learning, from structured or unstructured datasets including images, videos, text corpora, etc., the methods used require a principle of explicability (Vuarin & Steyer, 2023), on pain of reproducing biases and stereotypes.

The potential opened up by AI software development is directly challenging practices and models in the information and documentation sectors. Information production, processing, analysis and communication functions are being reinvented to take account of the possibilities offered by this type of technology.

After an initial phase of discovery, the various players in the information-communication field are rapidly moving into a phase of operational appropriation of these software tools in their business processes (Chartron, Raulin, 2022), (Tarride, S., Boillet, M., Moufflet, 2023), (Mignot, L. & Schultz, É. (2022), (De Corbière & all, 2023).

The speed of appropriation of this new info-communication environment leads us to project the next Document et Société conference in September 2024 in a context of practical implementation of this generation of tools for different sectors of activity.

This call for papers therefore invites us to analyze and explain the changes in practices, the new man-IA arrangements and the regulations at work in different fields of activity. How are we building forms of co-design situated with these AI bricks that are inserted into the majority of human activities as well as in the majority of professions (Condé, Ferguson, 2023)? Expected contributions may propose the following approaches:

- Analysis of the structuring of AI-based document ecosystems,

- Reflective feedback on the deployment of such tools,

- Observations of localized uses,

- Co-design modalities between these technologies and human expertise,

- Regulations found by the professions,

- Renewed interactions and skills associated with Prompting and Fine Tuning in particular,

- The biases identified in deep learning and the behaviors adopted as a result,

- The impact on organizations and the recognition of authoriality,

- The legal and economic issues associated with these technologies in the context of a renegotiated value chain,

- Governance and the implementation of regulations at different scales (national, international and European) aimed at containing the uses of AI.

Contributions are expected in the areas of press and media, documentation, publishing and libraries, economic intelligence and business intelligence, and teaching and education.

Axis 1: Press and media

Artificial intelligence represents a major challenge for the press and media on several levels (Dierickx, 2019). The automatic generation of content from vast data sets is fundamentally transforming professions. Whether producing news, financial reports or data summaries, these transformations are replacing journalists with more complex tasks requiring analysis and a human perspective. The personalization of the user experience by AI algorithms can also improve reader engagement in increased loyalty. To what extent are these two powerful developments overturning the skills expected of the field and the place of professionals in information production?

The use of AI in the media also raises ethical questions, such as the responsibility for information generated by algorithms and the protection of users' privacy. At the same time, automated fact-checking solutions can be used to quickly verify the veracity of information, helping to combat the spread of fake news. AI systems can analyze large quantities of data to detect inconsistencies and potential lies in articles. Against this backdrop, how do the media build trust with society?

Economically, cost reduction and increased efficiency are also driving changes in press and media models. By automating certain tasks, the media can reduce their operating costs, but for what new balance?

Area 2: Documentation, libraries and publishing

In the documentation and library sector, AI represents a significant challenge across the entire document lifecycle: production, indexing, classification, analysis and restitution (Hussain, 2023)... Moreover, conversational modules are being grafted onto information search engines, so won't document searching give way to a refined conversation with increasingly specialized generative AI? Interactions are thus being reshaped, with the Boolean equation giving way to the art of Prompting and Fine tuning. What skills do we need to develop in this new context? Are we witnessing the development of purely heuristic and pragmatic approaches? Or do these new forms of interaction call for deeper knowledge?

 As for publishing, new forms are emerging, combining humans and generative AI in augmented creativity. But where does the novelty lie in relation to the already numerous experiments in new digital editorial forms? In the context of professional publishing, particularly in the legal field, platforms are gradually being enhanced with new functionalities based on deep learning for highly specialized content. What is the feedback from users? Where does human expertise fit in, and in what new roles?

How are professionals positioned in this new environment? On processing and enrichment missions, how can we assess the potential added value of AI-based processing and its place in documentary processes?

How do the browsing and search interfaces offered in libraries today compare with interfaces that make extensive use of AI tools to optimize documentary responses?

Area 3: Watch, economic intelligence and business intelligence

Generally speaking, AI-related technologies have gradually been integrated into business intelligence platforms, which - let's not forget - are primarily designed to gather strategic information for a pre-defined environment. All stages of the intelligence process are involved: web search modules that harvest a range of sources, document processing and analysis aids. Automation is now amplified with the processing of multiple documents, the identification of sources in a corpus, the detection of any AI used for editing, and so on.

From intelligence to cyber-surveillance, there was only one step to take, and many players have taken it (e.g. Amisoftware and Bertin IT), just as the exploitation of personal data has become an infinite manna for high-tech companies. It would therefore be useful to look at the shifts in skills and the objectives pursued by business intelligence when its players use AI.

BI platforms that thrive on dashboard management offer companies new investigative tools capable of mapping themes, automatically generating recommendations for reporting, predicting behaviors and anticipating evolutions. Will this transform the intelligence, information intelligence and BI professions? And aren't data processing skills being over-valued, to the detriment of communication skills?

Axis 4: Teaching and education

The challenges and uses of AI technologies in teaching and education are opening up lively debates at both university and secondary school level. Should these technologies be rejected or integrated into learning processes? How should they be supervised? Can pedagogical and educational models and objectives (development of reasoning and critical thinking) be reconciled with the logic of individualized performance that underlies many algorithmic projects in this sector, such as individualized coaching? What help can teachers find in appropriating AI bricks? What skills are required? How can the educational institution support them (Unesco, 2023) (Colin de la Higuera, Jotsna Iyer)? At university, uses are developing rapidly for both teachers and students (Compilatio, Sphinx, 2023). This raises the question of the types of use permitted, their transparency and renewed evaluation methods in a context where writing a dissertation is no longer a guarantee of personal work.

Following on from the 6th Document numérique et Société conference on Big Data and Open Data, organized by ESI in Rabat, and the 7th conference on Humans and Data, organized by Crem in Nancy, this new conference aims to address AI in the information, press and media, library and publishing, documentation and intelligence sectors. It will encourage case studies, observations and feedback, as well as theoretical and critical reflection.

The Document Numérique & Société (doc-soc) conference has been held every two years since 2006, bringing together researchers interested in the evolution of the place of the document in our society, and shedding light on the economic and societal dimensions of documents and digital data.

Bibliography 

Broudoux, E. (2022). Éditorialisation et autorité. Dispositifs info-communicationnels numériques.De Boeck Supérieur, Information et Stratégie, 224, 978-2-8073-4084-8.⟨hal-03942725⟩

Chartron, G., Broudoux, E. (2015). Big Data - Open Data : Quelles valeurs ? Quels enjeux ? Actes du colloque « Document numérique et société », Rabat. De Boeck Supérieur.https://doi-org/10.3917/dbu.chron.2015.01

Chartron, G. & Raulin, A. (2022). L’intelligence artificielle dans le secteur de l’information et de la documentation : défis, impacts et perspectives. I2D - Information, données & documents, 1, 8-12. https://doi-org.proxybib-pp.cnam.fr/10.3917/i2d.221.0008

Colin de la Higuera et Jotsna Iyer, L'IA pour les enseignants, un manuel ouvert,

https://www.ai4t.eu/book/ia-pour-les-enseignants--un-manuel-ouvert-1/about-this-book?path=index

Compilatio, Sphinx, Résultats d’enquête : enseignants et étudiants confrontent leurs regards sur l’IA, 7/11/23, https://www.compilatio.net/blog/communique-presse-enquete-ia-2023

Condé, Ferguson (2023), rapport d’enquête usages et impacts de l’IA sur le travail, au prisme des décideurs,  https://travail-emploi.gouv.fr/IMG/pdf/enquete_laboria.pdf

De Corbière, F., Elie-Dit-Cosaque, C. & Leclercq-Vandelannoitte, A. (2023). Intelligence artificielle et recherche en management des systèmes d’information : menace ou opportunité ?. Systèmes d'information & management, 28, 3-9.

https://doi-org.proxybib-pp.cnam.fr/10.54695/sim.28.1.0003

Dierickx, L. (2019). Information automatisée et nouveaux acteurs des processus journalistiques. Sur Le Journalisme, About Journalism, Sobre Jornalismo, 8(2), 154–167. https://doi.org/10.25200/SLJ.v8.n2.2019.408

Epron, B., Vitali-Rosati, M., L’édition à l’ère numérique. La Découverte, « Repères », 2018, ISBN : 9782707199355. DOI : 10.3917/dec.epron.2018.01. URL : https://www.cairn.info/l-edition-a-l-ere-numerique--9782707199355.htm

Hilt, K. (2017). What Does the Future Hold for the Law Librarian in the Advent of Artificial Intelligence? / Que réserve l’avenir pour le bibliothécaire de droit avec la venue de l’intelligence artificielle?Canadian Journal of Information and Library Science 41(3), 211-227. https://www.muse.jhu.edu/article/686190

Hussain, A. (2023), « Use of artificial intelligence in the library services: prospects and challenges », Library Hi Tech News, Vol. 40 No. 2, pp. 15-17. https://doi.org/10.1108/LHTN-11-2022-0125

Lelu, A., & Rosenblatt, D. (1986). Représentation et parcours d'un espace documentaire. Analyse des données, réseaux neuronaux, et banques d'images. Les Cahiers de l'Analyse des données, 11(4), 453-470.

LINC-CNIL, [Dossier IA générative] - ChatGPT : un beau parleur bien entraîné, avril 2023,https://linc.cnil.fr/dossier-ia-generative-chatgpt-un-beau-parleur-bien-entraine

Mignot, L. & Schultz, É. (2022). Les innovations d’intelligence artificielle en radiologie à l’épreuve des régulations du système de santé. Réseaux, 232-233, 65-97. https://doi-org.proxybib-pp.cnam.fr/10.3917/res.232.0065

Tarride, S., Boillet, M., Moufflet, JF., Kermorvant, C. (2023). SIMARA: A Database for Key-Value Information Extraction from Full-Page Handwritten Documents. In: Fink, G.A., Jain, R., Kise, K., Zanibbi, R. (eds) Document Analysis and Recognition - ICDAR 2023. ICDAR 2023. Lecture Notes in Computer Science, vol 14189. Springer, Cham. https://doi.org/10.1007/978-3-031-41682-8_26

Unesco (2023), Guidance for generative AI in education and research, 44p,https://www.unesco.org/en/articles/guidance-generative-ai-education-and-research

Vuarin, L. & Steyer, V. (2023). Le principe d’explicabilité de l’IA et son application dans les organisations. Réseaux, 240, 179-210.https://doi-org/10.3917/res.240.0179

Authors are invited to submit their proposals online in the form of an anonymized .doc or .rtf. or .odt file (double-blind evaluation) via the platform

https://docsoc2024.sciencesconf.org/ 

Paper proposals should respect the following format: around 5000 signs, (spaces included, plus bibliography), presentation of the problem, theoretical framework, with central bibliographical references and, as appropriate, presentation of methodology and data. Papers in English are accepted. Bibliographical information must accompany the proposal.

Proposals must be submitted on the SciencesConf.org platform no later than. The papers resulting from the selected proposals, after evaluation, are to be presented at the conference on 2024. The proceedings will be published after the conference in De Boeck's "Information & Stratégie" collection, in early 2025.

Timetable

Call for papers: December 4, 2023

Deadline for submission of abstracts: February 8, 2024

Extended deadline for abstracts : 26 février 2024

Scientific committee evaluations: March 10, 2024

Notification of responses to authors: March 29, 2024

Final papers expected: May 31, 2024