AI and Media
WASP-HS Community Reference Meetings (CRMs) are dedicated for public and private organizations in Sweden for the purpose to learn about challenges and questions of their interest, and for WASP-HS to share recent research development within the program in order to identify opportunities for collaboration in different sectors.
This CRM is in collaboration with WASP Research Arenas – Media and Language.
It is not without reason that our current age is sometimes referred to as the media age. According to Mediebarometern 2020, the average Swede consumes 7 hours of media daily. The most frequently accessed media types are radio, tv, music, and social media (video games were not included in the study). At the same time, media lends itself well to AI applications: There is a huge abundance of digital data, a large part of which consists of text and images, two modalities where generative models such as GPT3 and DALLE have proved particularly successful. There are also strong commercial interests to motivate investments in new products and formats. It stands to reason then, that if AI comes to be widely adopted by the Media industry, and we spend almost half of our waking time on Media, then advances in Media AI are likely to have a pronounced impact on our daily lives.
In this Community Reference Meeting, we discuss how AI is being put to use in different media sectors, what helps and hinders the technological uptake and what are potential consequences for people and for society. Similar to previous CRM’s, a number of themes will be discussed in parallel at different roundtables. New this time is that the event will be hybrid, with discussion tables spread over three locations: At Chalmers in Gothenburg, we will look at AI for Text Generation, and How AI Transforms the Production and Consumption of Arts, Music, and Media. At Electronic Arts / DICE Office in Stockholm, the themes are AI for Gaming, AI for Movie Production, and AI for Music. At Umeå University, finally, the focus is AI in Politics. In these discussions, we will ask questions such as how game development can be made more efficient (and fun) through AI-driven testing, and how AI-driven political advertising can come to interfere with our democratic elections. As usual, the outcomes of the discussions will be summarised and made available to the community through the WASP-HS report series. The event kicks off with a keynote by Mark Harrisson, CEO, the Digital Production Partnership, which will be broadcasted live to all locations.
Roundtables are happening on-site at Umeå University, Electronic Arts / DICE Office in Stockholm, and Chalmers University of Technology in Gothenburg. However, there will be an opportunity to listen to the keynote and the reflections from the roundtables online via Zoom for those who whishes to do so.
13.10-13.30 Keynote – The Impact of AI on the Broadcasting Industry
Mark Harrisson, CEO at the Digital Production Partnership
13.30-15.00 Roundtable Discussions
in Umeå, Stockholm, and Gothenburg
15.00-15.30 Reflections from the roundtables
A Shift in Culture Through AI: How Does AI Transform the Production and Consumption of Arts, Music, and Media, located at Chalmers University Campus, Gothenburg
Chair: Kivanc Tatar, Assistant Professor, Interaction Design and Software Engineering division, Department of Computer Science and Engineering, Chalmers University of Technology
Co-Chair: Petter Ericson, postdoctoral fellow, Department of Computing Science, Umeå University
Increasing ubiquity of computing resources has enabled many to try, run, and observe AI technologies for music, visual arts, and media. The lines and borders between the audience and the artist are becoming less clear as AI content generators have become more accessible — as simple web pages for novice users, open-source code, and open-access models for experts. The explosion of content generated by such AI models has initiated a cultural shift in arts, music, and media, where roles are changing, values are shifting, and conventions are challenged. The readily available, vast dataset of the internet has created an environment for AI models to be trained on any content on the web, whether that content is collected by scraping the web, or via ready-made datasets and APIs provided freely by researchers, organisations and practitioners. With AI content generator models shared openly, and used by many, globally, how does this new paradigm shift challenge the status-quo in artistic practices? What kind of changes will AI technology bring into music, arts, and new media?
AI for Text Generation, located at Chalmers University Campus, Gothenburg
Chair: Aarne Ranta, Professor, Computing Science Division, Department of Computer Science and Engineering, Chalmers University of Technology
Automatic text generation provides a cheap and fast way to create content in natural language. Recent systems based on machine learning, such as GPT-3 and BERT, can generate text that is fluent but whose content is difficult to control. In contrast to this, classical rule-based methods, such as text robots, can be reliable in rendering content, but often in a clumsy and repetitive style. The round table will gather researchers developing methods of both kinds, as well as users of text generation from industry and society. We will discuss the choice of optimal methods for different use cases, as well as ideas for combining them in fruitful ways.
AI for Music, located at Electronic Arts / DICE Office, Stockholm
Chair: Bob Sturm, Associate Professor, Division of Speech, Music and Hearing, Royal Institute of Technology
Co-Chair: Andre Holzapfel, Associate Professor, Division of Media Technology and Interaction Design he KTH Royal Institute of Technology
This roundtable will focus on the involvement of artificial intelligence (AI) in a variety of music practices, from listening, to composition and performance, to analysis and criticism. There exist many real world commercial examples of these, from music and media recommendation by Spotify and YouTube, to music generation by jukebox and boomy.com. There are also a variety of niche interests, such as traditional Irish music identification, folk tune generation, Baroque four-part voice leading, or endless metal music generated by deep neural networks. What kinds of impacts do these applications have? What legal and ethical considerations should be made by users and developers of these applications? How should the values of music communities be considered in the development and application of AI for music?
AI for Gaming, located at Electronic Arts / DICE Office, Stockholm
Chair: Konrad Tollmar, Electronic Arts and Royal Institute of Technology
This roundtable will discuss Game AI and the opportunities and challenges in using AI in tomorrow’s games. Game AI in today’s game could be somewhat deceptive. Most characters and interactive worlds in computer games today, like the enemies in a first-person-shooter, or the Sims are not examples of artificial intelligence. They’re scripted AI’s even if they provide the perception of smart behavior on the screen. However, they don’t learn or develop new behavior over time as we imagine in today’s AI research. What will the future of games driven by modern AI become like? We see clear trends in using AI for content generation, driving gameplay, and facilitating social media experiences around games. We have the Metaverse, VR and AR a bit further down the line where the boundaries between digital and physical play blur out and get mashup. “The biggest challenge for AI is to mimic what is perhaps the most complex and mysterious capacity of the human brain: Imagination,” says Julien Desaulniers, the programming team lead of AI and Gameplay on Assassin’s Creed Valhalla. There are also substantial legal and ethical considerations of using AI in digital games. How to address these?
AI for Movie Production, located at Electronic Arts / DICE Office, Stockholm
Chair: Mark Harrisson, CEO at the DPP
AI in its many incarnations – speech recognition, object detection, video summarisation, etc. – has come and gone in turns from the broadcasting industry’s hype radar. A few years back, it was high up on the DPP Future Prediction list, but is nowhere to be seen in this year’s list. There seems to be a widespread disillusionment in the industry as to what AI has promised (or rather, what has been promised by unserious actors on behalf of AI), compared to the actual impact that it has delivered. At the same time, AI research is progressing at a break-neck speed, with generative models such as GPT3 and DALLE pushing the limits of what is possible with in terms of content generation. Why then are these solutions so slow to make it onto the market? Is technology transfer particularly difficult in the broadcasting industry, or is the problem more general? And could it be that the challenge is not the technology itself, but the communication around it, and adapting organizational patterns? To guide us in these discussions, we take help from Mark Harrisson, CEO of the UK-business organization the Digital Production Partnership, who has decades of experience working with technology transfer in the media industry.
AI in Politics, located at Umeå University Campus
Chair: Simon Lindgren, Professor, Department of Sociology, Umeå University
Co-Chair: Mona Forsman, Head of the Research and Development at Adlede
This roundtable addresses issues that relate to the use of AI-driven automation in political contexts online. Both in terms of strategic and large-scale campaigns by parties and governments, and from a more grassroots-oriented perspective, focusing for example on hacktivist uses of social bots. Both top-down and bottom-up interventions raise ethical issues. Given the new technological opportunities, do we need new standards or legal frameworks to increase transparency and protect privacy, or is self-regulation enough? Where should we draw the line between information and manipulation, or between participation and hacktivism?
Please note that the event will take place in English only. More information about location and address will be given closer to the event to those who register. The numbers of participants is limited and the registration can close before the registration deadline on October 7 if the limit of registrations is reached.
Registration is closed. See other events.
For further questions regarding the event, please contact our event coordinator.