MEDIA INDIGENA 285
Added 2022-04-10 01:00:15 +0000 UTCMI on A.I. / MI 285
ON THIS WEEK'S INDIGENOUS ROUNDTABLE:
Getting real with artifical intelligence. Hardly a day goes by it seems without news of some ‘revolutionary’ A.I.-driven tool ushering in a brave new world. Less said is who’ll be left out or left behind. Which is why, when it comes to Indigenous content, some fear much of artificial intelligence remains superficial ignorance. But can ‘The Cloud’ incorporate culture? Can we 'Indigenize' as we digitize? And can the digital be made relational?
Joining host/producer Rick Harp to tangle with these tricky, trippy questions and more are Kim TallBear, professor in the Faculty of Native Studies at the University of Alberta, and Trina Roache, Rogers Chair in Journalism at the University of King’s College.
. . . . . . . .
LINKS REFERENCED / CONSULTED THIS EPISODE:
• “Artificial Intelligence (AI)” IBM Cloud Hub
• Work Without the Worker: Labour in the Age of Platform Capitalism Verso Books
• "Microworkers are 'Disempowered to a Degree Previously Unseen in Capitalist History'" Jacobin
• "Making Kin with the Machines" Journal of Design and Science (JoDS)
• “How Native Americans Are Trying to Debug A.I.’s Biases” New York Times
• PHOTO: Red School House American Indian Movement Interpretive Center
• Intelligent Voices of Wisdom IVOW.ai
• VIDEO: Sina Storyteller Demo Vimeo
. . . . . . . . .
LISTEN NOW:
https://mediaindigena.libsyn.com/getting-real-with-artificial-intelligence-ep-285
Comments
Elizabeth, thank you so much for your generous response: lots to chew on here! :D
Rick Harp
2022-04-27 00:36:41 +0000 UTCHi! So I do relevant research to this episode - I'm a (settler) sociologist who studies hegemony in computer science education. Since you had questions in the episode: an algorithm is generally seen as any formal procedure for computing something. So for example, timekeeping methods (calendars & their calculations) are seen as algorithms for keeping track of time. Patterns for weaving are also considered algorithms. As are protocols for deciding how many fish to harvest in order to keep populations stable. There's a whole (niche) field called ethnocomputing which looks at traditional algorithms used by cultures around the world. Ron Eglash's research group has been teaching culturally-responsive computing based in ethnocomputing for quite some time now, here's an overview of some of their projects with First Nations: Eglash, R. (2007). Ethnocomputing with native American design. In Information technology and indigenous people (pp. 210-219). IGI Global. https://drive.google.com/file/d/1tOUfm9NLxdCihUj2f0ATvUakEGpwr4L8/view?usp=sharing In the episode you asked if language would be considered an algorithm. I would say as somebody with multiple computer science degrees that typically language is seen as the vehicle for how you communicate/record an algorithm. Language is so rich and multifunctional that it seems like much more than an algorithm to me! And indeed in theory of computation, language is seen as more akin to computers (ie that which is able to do any given computational work - human or machine) than the algorithmic input provided. (Note: "computer" used to mean a human who did calculations for a living - you'll see the term "computing machine" used in CS as necessary to differentiate an abstract computer from a physical computing machine.) Okay, moving on to the question of what an Indigenous computing would look like. Lisa Nathan's group has been thinking about this as settlers and not pretending to have an answer but their attempts are helpful - here's such a paper I personally assign in my own teaching, and has a bunch of helpful references: Nathan, L. P., Kaczmarek, M., Castor, M., Cheng, S., & Mann, R. (2017, June). Good for whom? Unsettling research practice. In Proceedings of the 8th International Conference on Communities and Technologies (pp. 290-297). https://open.library.ubc.ca/collections/52383/52383/items/1.0398209 Typically in the tech world a piece of technology is valued based on its performance - whether it's faster than comparable technologies, whether it's more accurate/precise, etc. But in the article Lisa Nathan et al argue for a value system where the relationships between designer and community should be how a piece of technology is valued, and how their experiences as settler designers working with Musqueam and other Indigenous groups changed their own approaches to computing research. Another group of people to know about is the critical curation studies folks. They've been thinking for a long time about questions of who has data, how data should be managed, etc. Data trusts are a thing they talk about. E.g.: Kukutai, T., & Taylor, J. (2016). Indigenous data sovereignty: Toward an agenda. ANU press. https://library.oapen.org/bitstream/handle/20.500.12657/31875/624262.pdf?sequence Since your episode was on AI & Machine Learning I think it's important to note there has been Indigenous-led effort for some time now to think about Indigenizing statistics. This is important because machine learning - the most popular tool for AI - is really just using statistics for fancy pattern matching. There's a whole book on Indigenous statistics worth having a look at: Walter, M., & Andersen, C. (2016). Indigenous statistics: A quantitative research methodology. Routledge. (I can't find a free copy of the whole thing but here's a chapter from it on good numbers: https://drive.google.com/file/d/1sE0kfAi3TsN7h-sP-5dR-jxSkYmmqULW/view?usp=sharing ) And there are also scholars rethinking/exposing relationships between Indigenous peoples and the big data used in machine learning. For example this paper by Radin on how the Pima Indian Diabetes Dataset has been used in ML: Radin, J. (2017). “Digital natives”: How medical and indigenous histories matter for big data. Osiris, 32(1), 43-64. https://www.journals.uchicago.edu/doi/pdf/10.1086/693853 I liked in your intro that you noted the difference between weak AI and strong AI. One thing that might be useful to know is that many computer scientists (myself included) are quite skeptical that strong AI may ever be possible. Strong AI has been promised as being "5 years away" for decades now. It seeems to be a perpetual hype that never dies. On top of the repeated hype part of why I'm personally skeptical is we finally seem to have hit the physical limit for how fast computers can be. For a long time chips got smaller every year and thus faster (since speed of light is a limitation), but now the chips are so small quantum effects are limiting them getting any smaller & faster. Finally, on the note of AIs as kin: the podcast Flashforward did an absolutely amazing job of tackling the state of robotics as it is now in their last four episodes. All four episodes are great and very educational (I really liked the one on sex bots because it was a take that was totally unlike all the moral panic takes I'm used to seeing - instead focusing the questions like what materials would be needed, like the issues of lithium mining, and so forth). But what's most relevant to the kin issue is the episode on robots and animals: ROBOTS: Could Robots Replace Animals? https://www.flashforwardpod.com/2021/12/21/final-episode-robots-could-robots-replace-animals/ In the episode Rose Eveleth talks to researchers on the history of how technology replaces labour and highlights how so much of the time it's not actually human labour that is replaced by machines, but animal labour. Horses are replaced by engines (still measured in horsepower), oxen by tractors, ferrets by little robots, etc. I'd definitely give it a listen for important context when it comes to the Making Kin with the Machines! Hope this is helpful for a follow-up episode. I hope all the links work but let me know if you hit paywalls!
Elizabeth Patitsas
2022-04-25 04:59:09 +0000 UTCThank you thank you thank you for this excellent episode. So important now and these need to be mainstream conversations or at a minimum as part of the House of Commons ETHI committee study on facial recognition.
LegallyAbigail
2022-04-10 15:07:30 +0000 UTC
