Can computers be feminist
- 1. can computers be feminist? procedural politics and computational creativity Gillian Smith Northeastern University e: gi.smith@neu.edu t: @gillianmsmith
- 2. agenda ➤ what is my background? ➤ intro to computational creativity and games ➤ building computational models ➤ considerations for procedural politics ➤ towards feminist computational models
- 3. about me ➤ Assistant Professor, Northeastern University ➤ Playable Innovative Technologies ➤ Art+Design, Computer Science ➤ Research in… ➤ computational creativity ➤ experimental game design
- 4. computational creativity designing software that (semi-)autonomously creates art hoopla collaboration viv
- 5. game design creating new kinds of playable experiences eBee threadsteading GrACE
- 7. computational creativity and formal modeling ➤ specification of desirable content, attained via: ➤ underlying data ➤ algorithmic structure ➤ art as expression of societal values ➤ notions of human creativity ➤ embedded meaning ➤ novelty and value ➤ historical, cultural context
- 8. knowledge representation what kind of data and at what level of granularity @thetinygallery (emma winston) spelunky (derek yu) mad libs
- 9. algorithm design capturing the (human?) creative process @DeepForger (alex champandard) refraction (adam smith et al.) a rogue dream (mike cook)
- 10. nature of the artifact formal theory of space of all potential pieces plotto (william wallace cook) baldur’s gate (black isle) launchpad (gillian smith et al.)
- 11. games and formal modeling ➤ playable formal models ➤ designer specifies data, rules ➤ system dynamics + player introduces emergence ➤ specification of: ➤ structure of people and society ➤ social interaction ➤ additional domains…
- 12. game design expressing social and political behavior as playable systems prom week (mccoy et al.) mass effect 3 (bioware) alice and kev (robin burkinshaw)
- 13. commitments ➤ code ➤ data ➤ framing
- 15. “The moral values we treasure are reflected in the beauty and truth that is emotionally transmitted through the arts. The arts say something about us to future generations. Ann P. Kahn
- 16. what’s in a name? ➤ phenomenon of creative AI systems with feminine names ➤ ELIZA ➤ ANGELINA ➤ DARCI ➤ Viv ➤ Siri… ➤ what is the assumed competence? ➤ how is their role traditionally gendered?
- 17. complexity of authorship and labor ➤ entwined relationships ➤ additional stakeholders? ➤ who receives credit? ➤ who avoids blame? ➤ who is paid? ➤ whose work is alleviated? ➤ whose work is enabled? developer software player/user creates generative model reflects on output synthesizes input generates novelty creates retellings curates artifacts shares to community
- 18. embedded biases ➤ generative systems produce a specification of sample artifacts ➤ explicit ➤ implicit ➤ what are the constituent parts? ➤ what are the probabilities they’ll be chosen? seen? ➤ how is information organized? www.icongenerators.net
- 19. cultural knowledge ➤ human creatives act within their own cultural context ➤ software alternatives ➤ crowdsourcing ➤ “expert” system ➤ what is ground truth? ➤ what is subjective truth? ➤ how is it contextualized?
- 20. rationality and logic ➤ Adam’s critique of “universal” logic ➤ do logic and rationality look the same across cultures? ➤ notion of machine as “unbiased” ➤ use randomness as a check on authorial bias Adam, A. 2005. “Knowing Subjects: AI from Feminist Philosophy.” In Mechanical Bodies, Computations Minds: Artificial Intelligence from Automata to Cyborgs, edited by Stefano Franchi, and Güven Güzeldere., 327–344. Cambridge: MIT Press.
- 21. towards feminist formal models a set of provocations
- 22. “I ask what it might mean to design— from their very conception—digital tools and applications that emerge from the concerns of cultural theory and, in particular, from a feminist concern of difference. -Tara McPherson, “Designing for Difference”
- 23. Courtney Toder, work-in-progress software should support richer identity models
- 24. we need to enable more diverse authors
- 25. need to surface bias implicit in algorithms and data
- 26. content filtering is insufficient
- 27. we escape culpability because software is “rational”
- 28. the fundamental flaw of big data is a lack of empathy
- 29. with thanks to… ➤ Amanda Phillips, Georgetown University ➤ Mike Cook, Falmouth University ➤ Tanya Short, Kitfox Games ➤ Mitu Khandaker-Kokoris, New York University ➤ Vi Hart, eleVR ➤ Marc ten Bosch, independent game developer ➤ Schloss Dagstuhl ➤ Banff International Research Station Gillian Smith Northeastern University e: gi.smith@neu.edu t: @gillianmsmith