Mapping Mixed-Initiative Creative Interfaces: How Can Humans and AI Collaborate in Creative Practice? - project 5
As part of the current bloom of artificial intelligence, we are seeing the rise of ‘creative AI’ – artificial intelligence that produces or supports creative work like making games, writing prose, composing music, or editing films. A particularly novel and interesting approach in this field are so-called Mixed-Initiative Creative Interfaces or MICIs (Deterding et al., 2017).
Rather than human creators using computers as a tool, or computers acting as creators on their own, in MICIs, human and AI engage in a creative dialogue, taking turns constraining, suggesting, producing, modifying, selecting, and evaluating creative outputs. Examples are systems like TANAGRA, where a human game designer suggests a rough draft of a game level, and the AI produces a detailed game map, automatically analyses whether the map is actually playable, and if not, suggests edits. Beyond games, we find MICIs in product and web design, painting, music, writing and story creation, film editing, etc.
The goal of this project is to map the design space of how humans and AI can collaborate in a creative endeavor. Using standard taxonomies of creative practices (framing, ideating, selecting, assessing, modifying, ...), you will collect existing MICI systems across domains (games, visual art, ...) and build a digital map which kinds of collaboration have already been realised in which domains and which therefore remain underexplored.
To do so, you will
- Put together and run an online survey asking people in industry and academia to submit MICIs they have developed;
- learn to do a systematic literature review on MICIs;
- build a taxonomy or map of all discovered MICIs;
- build and fill a database and website showing MICIs, similar to popular sites lile http://www.creativeai.net and https://www.microsoft.com/buxtoncollection/browse.aspx
Number of places available
We are looking for two advanced undergraduate or master's students interested in creative AI.
Students should be able to do a systematic literature review involving searching, filtering, and coding literature; designing and monitoring an online survey design; and ideally, building a web collection using existing open source frameworks like http://www.telescopeapp.org/.
Dr Sebastian Deterding
Dr Jon Hook
Deterding, C. S., Hook, J. D., Fiebrink, R., Gow, J., Akten, M., Smith, G., ... & Compton, K. (2017). Mixed-Initiative Creative Interfaces. In Proceedings of CHI 2017 Extended Abstracts. ACM.
How to Apply
For more details on the summer school application process (including eligibility and funding) please see the overview page at: http://www.digitalcreativity.ac.uk/dclabs-summer-school-scholarships-2017