Ep. #40 Robot Reflections: Inger Mewburn on researching researchers & welcoming our robot overlords

“Machine learning is a broad area of study, and that’s one thing you don’t see from a distance, is how broad it is. We like to do what we call human-in-the-loop type of machine learning, which is a co-creation of something. So, what we’re using is the machine to, sort of, both capture our knowledge and reflect it back at us, but at the same time the machine’s training us … When a social scientist or an anthropologist or anyone for that matter, who’s looking to collaborate with a machine, it’s a collaboration: they shape you, you shape them, what you make is this nexus in between.”

In episode 5 of our STS Series, Inger Mewburn, Associate Professor and Director of Research Training at the Australian National University, founder of the popular blog The Thesis Whisperer and author of How to Tame Your PhD, Becoming an Academic, How To Be An Academic, and How to Fix Your Academic Writing Trouble, chats to our own Jodie-Lee Trembath. They talk about what it means to be a post-disciplinary or interdisciplinary scholar using ethnography to understand social phenomena, about machine learning and the values that are reflected through the machines we create, and how sometimes these values are ones that we don’t really want to see, and about the future of research and work where algorithms and technology form a collaborative effort between humans and robots.

QUOTES

“Ethnography is just a really considered look at what’s right in front of your face – the familiar strange, right. As soon as you slow things down, start to take things apart, you notice things that, in the moment, just pass you by.”

“I hate writing academic papers…I hate it with the heat of 1000 blazing suns…it’s not that I can’t do them, I find the format really boring, really stifling – the actual writing process is dull. I find the circulation of them to not be circulating to the people that they actually effect… my work helps supervisors and students, most of the time they don’t read [that] research in a research journal.”

“When it comes to translating human jobs into computers, I think I’d rather have that done by people like anthropologists – no disrespect to people with business degrees, but they’re not trained to look for the same things and to think holistically, the way that an anthropologist thinks”

“One of my queer friends said to me once that every person who is non-binary, or an othered sexuality is a natural anthropologist. He said because you grow up sitting on the sidelines and trying to work out how you do or don’t fit.”

“I think we can make really terrible machines, scary, well consider medicine – medical anthropology, right … machine learning in health, machines can look at, say, a whole lot of your test results and spot your risk factor for something, or they could spot your chance of surviving given a certain kind of treatment and say ‘well actually it’s not worth it’. So, if a machine like that’s embedded where it’s upfront and you can see it and you can say, ‘well, actually, bugger you machine, we’re giving this person the cancer treatment because it will prolong their life’ … but what if that’s not visible? What if [the machine] just sits behind, deciding what drugs get dispensed or not and you don’t see the basis of that decision-making in action? That decision making is invisible to you. An anthropologist would think ‘what’s that effect on community? What’s the kinship networks? What’s the broader societal impact of not giving this person another 3-6 months?’ That’s what an anthropologist would think.”