Make me a video meta
Abstract
Unfortunately, we have had to cancel this seminar. We hope to arrange it again for next semester.
Many will be aware of the explosion of (alleged) creativity going on right now in the digital space, as tools like DALL-E2, Stable Diffusion and Meta’s “Make-a-Video” allow anyone, however unartistic or untrained, to create images and even videos using only textual prompts . Before that, a similar wave of amazement and innovation followed the announcement by Open AI, and then many others, of “large language models” such as GPT-3 which can generate convincingly human-appearing text from prompts, and are now already being used commercially in eg chatbots and journalism.
The excitement inevitably comes alongside harms and risks. Is this Fake News 2.0? Do such models inevitably embody historic biases and failures to represent women and minorities? Will this be just a new way to sexually abuse and impersonate women online?
This talk is a meander through the large amount of issues emerging, will weave in the EU’s AI Act proposal for regulating AI generally (somehow), consider private ordering and technical means of governance, and generally attempt to sketch out a workspace in this area for the future.
All are welcome.
Refreshments will be available.