Artificial Intelligence

Machine Reflections – Workshop on Visual Generative AI, 11 July

Posted on    by Stefan Luca
Artificial IntelligenceBlogEvents

Machine Reflections – Workshop on Visual Generative AI, 11 July

By 11 July 2023February 4th, 2024No Comments

Join us in Glasgow for a networking workshop open to researchers across various disciplines (computer science & engineering, data science, digital arts, law, STS, etc.) addressing the challenges raised by new visual generative AI (VGAI) tools. This focus aims to be wide enough to capture the common technical underpinnings of image generation, while sufficiently narrow to address concerns around specific use cases, from journalism to book publishing, taking into account the structure and practices of different industries. This interdisciplinary encounter seeks to explore the interplay between VGAI tools and various creative ecosystems, drawing on CREATe’s expertise on digital copyright and content moderation issues.

The workshop is structured as a series of roundtables anchored by expert presentations from CREATe or invited speakers, addressing the following four themes:

  1. Creativity on notice

Prompted by recent efforts to allow artists to opt-out from large training sets, we will explore lessons that may be drawn from the empirical evidence on copyright mechanisms. We will further interrogate the technical and organisational dimensions of various enforcement approaches, such as watermarking, filtering, and notices.

  1. Artificial indignity and virtual representation

The circulation of AI-generated content raises a familiar set of trust and safety concerns. More specifically, the reception of VGAI tools has been affected by incidents of stereotypical and offensive representation of women and racialised people. Conversely, efforts to harness these tools for more diverse representation have been criticised for side-lining those whose inclusion is ostensibly promoted.

  1. Technical stack

The current configuration of the technical stack, from data collection, through model training, fine-tuning, and tool deployment, channels criticism and demands towards a small number of gate keepers. Nevertheless, lower barriers to entry are democratising access to these tools. This in turn requires re-thinking the distribution of responsibilities between the various levels of the technical stack and identifying opportunities for collaboration.

  1. Normative links

Building on new research spearheaded by Professor Lilian Edwards on the terms and conditions under which generative AI models are released, we will address normative developments prompted by VGAI, ranging from those embedded in APIs to social and professional norms emerging in various creative communities. The potential impact of adversarial, remedial, and subversive tools will also be considered.

Practical details

Workshop date & venue

Tuesday, 11 July, 9.00 – 13.00

Advanced Research Centre

University of Glasgow

To apply, send a 100-word bio and a 300-word summary of how your research addresses one the workshop’s themes to stefan.luca@glasgow.ac.uk, no later than 30 June. The deadline is extended to Monday, 3 July.

One night of accommodation and domestic travel to Glasgow will be covered. Prospective participants with care responsibilities may apply for a limited number of £100 contributions towards care costs.

Details about the programme and invited experts will follow soon, so check back regularly.

 

This event is made possible by a Turing Network Development Award.