Workshop date: March 22, 2024
Towards Ethical and Inclusive Conversational AI:
Language Attitudes, Linguistic Diversity, and Language Rights
(TEICAI)
Workshop at EACL 2024 (March 17 - 22, 2024, Malta)
Our vision
Conversational language technologies (chatbots, voice assistants, and multimodal conversational interfaces) are becoming increasingly complex and common in everyday life. Various language theories (such as speech act theory, politeness theory, conversation analysis, and interaction theory) have started influencing their development. At the same time, the development of these technologies is often driven by technology-related concerns and tends to overlook users’ needs and socio-cultural contexts. This, combined with the scarcity of human rights regulation of AI, raises concerns about linguistic discrimination, exclusion, surveillance, and security risks. In addition, training data for conversational AI mostly comes from written rather than interaction-based language data sets and often does not include gestural, social, and emotional aspects that are fundamental to human interaction. In the same vein, Sign Language is rarely facilitated. To promote a positive impact of conversational technology on linguistic diversity and inclusion, it is imperative to strike a balance between technological concerns and socially relevant matters.
Our workshop aims at addressing these issues by using a holistic approach that involves dialogue and collaboration among technologists, linguists, policymakers, and communities involved in the development and commissioning of conversational AI systems.
Workshop program
The workshop takes place on the 22. March 2024 in Corinthia Hotel (Ballroom: Bastion 1)
9:00 - 9:30 Opening and welcome by organisers
9:30 - 10:30 Keynote 'Culturally-aware educational language technologies' by Justine Cassell
10:30 - 11:00 Coffee break
11:00 - 12:00 Paper session 1
Why academia should cut back on general enthusiasm about CAs. Alessia Giulimondi (position paper).
Socio-cultural adapted chatbots: Harnessing Knowledge Graphs and Large Language Models for enhanced context awarenes. Jader Camboim de Sá, Dimitra Anastasiou, Marcos Da Silveira and Cédric Pruski (position paper).
Bridging the Language Gap: Integrating Language Variations into Conversational AI Agents for Enhanced User Engagement. Marcellus Amadeus, Jose Roberto Homeli da Silva and Joao Victor Pessoa Rocha (position paper).
Non-Referential Functions of Language in Social Agents: The Case of Social Proximity. Sviatlana Höhn (short paper).
12:00 - 13:45 Lunch break
13:45 - 14:45 Paper session 2
How Do Conversational Agents in Healthcare Impact on Patient Agency? Kerstin Denecke (long paper).
How should Conversational Agent systems respond to sexual harassment? Laura De Grazia, Alex Peiró Lilja, Mireia Farrús Cabeceran and Mariona Taulé (short paper).
Making a Long Story Short in Conversation Modeling. Yufei Tao, Tiernan Mines and Ameeta Agrawal (long paper).
14:45 - 15:30 Keynote talk 'Language technology for equality, diversity and inclusion' by Bharathi Raja Chakravarthi
15:30 - 16:00 Coffee break
16:00 - 16:40 Round-table discussion "Successful projects with mixed expertise in conversational AI, language rights and language ideologies". Speakers: Doris Dippold, Nina Hosseini-Kivanani, Justine Cassell, Valentina Pyatkin. Moderator: Sviatlana Höhn
16:40 - 16:50 Closing and farewell by organisers
Invited speakers
Justine Cassell
Directrice de Recherche, Inria Paris
Dean's Professor, Carnegie Mellon University
Culturally-aware educational language technologies
Children seamlessly shift their ways of speaking, adopting and adapting language they hear spoken by adults around them, as well as constructing their own variants. These speaking styles play an important role as children experiment with who they want to be, and how they want to be perceived. They also allow children growing up in situations where different dialects or languages are in contact to mark their affiliation to dominant and minority ethnic, racial, and gender identities.
Young people who move between marginalized and mainstream communities often report code-switching as a way to maintain affiliation with their home community, while also making their way in a world where the standard dialect is associated with various kinds of success. However, moving back and forth may lead to what Ogbu (2008) has called “oppositional culture” to describe how school systems may inadvertently set up a situation where the student feels the need to define her identity contra the expectations of the school, and for that reason to refuse the dialect that the school insists on.
Unfortunately, language technologies, including and very persistently, educational technologies, may inadvertently carry negative stereotypes about ethnicity and how it is carried in language, leading to stress and lack of a sense of agency as children try to navigate a path that allows them to benefit both from the support of their community and the opportunities offered by mainstream education.
My students and I have examined this issue by building “culturally-aware” educational language technologies, and specifically “virtual peers” that either speak only the child’s own dialect, or that model a code-switching strategy between what linguists call low-prestige and high-prestige dialects. Results using a variety of methodological approaches, in both one-shot and longitudinal studies, demonstrate the positive impact of technology such as these that take issues of culture, and of power, into account on children’s school performance. On the other hand, careful assessments of the children’s reactions to the technologies shows that they will need further development to improve the children’s own internalized biases against low-prestige dialect speakers.
Bharati Raja Chakravarthi
Assistant Professor at the National University of Ireland Galway
https://www.insight-centre.org/our-team/bharathi-raja/
Language technology for equality, diversity, inclusion
Equality, Diversity and Inclusion (EDI) is an important agenda across every field throughout the world. Language as a major part of communication should be inclusive and treat everyone with equality. Today’s large internet community uses language technology (LT) and has a direct impact on people across the globe. EDI is crucial to ensure everyone is valued and included, so it is necessary to build LT that serves this purpose. Recent results have shown that big data and deep learning are entrenching existing biases and that some algorithms are even naturally biased due to problems such as ‘regression to the mode’. Our focus is on creating LT that will be more inclusive of gender, racial, sexual orientation, persons with disability. Over the past few years, systems have been developed to control online content and eliminate abusive, offensive or hate speech content. However, people in power sometimes misuse this form of censorship to obstruct the democratic right of freedom of speech. Therefore, it is imperative that research should take a positive reinforcement approach towards online content that is encouraging, positive and supportive. Until now, most studies have focused on solving this problem of negativity in the English language, though the problem is much more than just harmful content. Furthermore, it is multilingual as well.
Organizers
LuxAI
University of Luxembourg
Luxembourg Institute of Science and Technology (LIST)
University College Dublin
State University of Moldova
University of Surrey
Zortify
Venue
TEICAI Workshop will take place at one of the venues of the EACL 2024: Corinthia St George’s Bay in Malta
Contact
General requests related to organisation and submission at TEICAI 2024: teicai2024@gmail.com
Where to find us?
Linkedin: TEICAI
Twitter: teicai2024