AI Sweden is looking for researchers in the humanities and social sciences as well as civil society organizations who are interested in joining a new interdisciplinary collaboration project, learning about generative language models (e.g. ChatGPT), and leveraging their expertise for the development of tomorrow’s AI language systems: the Interdisciplinary Expert Pool for Natural Language Understanding (NLU).

AI Sweden is the national center for applied AI in Sweden, funded by Vinnova and partners from the public and private sectors and academia. Together with RISE and WASP we are building generative language models for the Swedish and other Nordic languages,GPT-SW3. It is this model type that forms the basis for services such as the recently much discussed ChatGPT. However, at AI Sweden we focus particularly on Swedish and Nordic languages with the aim of building a model as representative as possible for use cases in Sweden - and we do so in a non-profit, transparent, and collaborative research environment.

Generative language models, such as ChatGPT and our own models GPT-SW3 are trained on large amounts of text data - so called training data (discussed in an article about training data for GPT-SW3). This training data determines how the model works and what kind of texts the model can generate. Choosing the “right” training data raises questions about representativeness, fairness, ethics, and equality. Examples of discriminating or in other ways “undesirable” AI systems have made clear that the development of these models as well as their application requires the involvement of a broad variety of knowledge domains and disciplinary backgrounds. Current debates about ChatGPT also make clear - once more - that we need to have informed debates about the cultural and social implications of these kinds of models and also ask big questions.

To tackle these challenges, we have started a new project at AI Sweden called “Interdisciplinary expert pool”. The idea of the project is to systematically involve scholars from the humanities and social sciences as well as civil society actors in the development of AI driven language technologies. The involved experts will at the same time get insights into how AI language technologies (e.g. ChatGPT) work and how they can leverage this new technology for qualitative and quantitative research in the humanities and social sciences.

Do you think this project sounds interesting and want to know more? Sign up for one of the info meetings which take place in week 10.

Please feel free to forward this email to colleagues who might be interested in the project and don’t hesitate to contact Francisca Hoyer (francisca.hoyer@ai.se) if you have any questions.