A Pilot Project in Open Scholarship
Rhetorical Depictions of Human Intelligence in Amazon’s Mechanical Turk
Amazon's Mechanical Turk (MTurk) is a crowdsource marketplace that allows individuals and organizations to solicit a distributed pool of workers for menial tasks that are impractical for computers. Examples of such "Human Intelligence Tasks" or "HITs" include identifying objects in images, correcting transcription errors, and verifying product descriptions. Although compensation for HITs is often substantially below minimum wage, the service is quite robust (Semuels, 2018; Hara et al., 2017; Cushing, 2012).
MTurk is active in several academic disciplines, both as a subject of study and—somewhat ambivalently—as a ready source of human research subjects (Hitlin, 2016; Hauser & Schwarz, 2015; Casler, Bickel, & Hackett, 2013; Barger, Behrend, Sharek, & Sinar, 2011; Buhrmester, Kwang, & Gosling, 2011; Paolacci, Chandler, & Ipeirotis, 2010). Although the service mostly consists of written text, there are few extant examinations of MTurk in rhetoric studies. This project seeks to help address this lack by collecting and analyzing a corpus of HIT descriptions to develop a preliminary understanding of how human intelligence is rhetorically depicted on this platform. This investigation will extend the disciplinary interrogation of contemporary digital literacy (Holmevik, 2012; Ulmer, 2003; Ong, 1982) by examining the role of attention in characteristic human labor. MTurk can provide valuable insight into related issues of identity and agency because it precisely separates people from machines through market forces that reveal what the former does more effectively or efficiently than the latter. Humans are often distinguished from non-human actors by virtue of an elevated intellectual or a moral trait that they possess, yet their characteristically mundane abilities potentially offer a more concrete and informative contrast. This study approaches distinctions between humans and machines as traces of evolving complex systems that contain various forms of agency (Cooper, 2010; Barad, 2007; Herndl & Licona, 2007; Miller, 2007). MTurk suggests how human and non-human agents may collaborate productively by performing specialized operations (Harman, 2008; Latour, 2005). This study thus eschews a Platonic view of humans as formerly apex cognitive agents who are being progressively displaced. Emerging from this study will be a functional model of contemporary human intelligence and agency articulated through language and legitimated by economic pragmatics.
The study's authors seek to add nuance to this project by enacting human/machine collaborations. The authors currently have only a general outline of the research project; they plan to hold scheduled live streaming video discussions to move through the composition process from initial idea to finished product. Viewers may participate in these discussions, which will be archived with associated materials—textual exchanges, research documents, associated online works, etc.—and publicly shared throughout the composition process. The article draft will be available online and open to public comments as it develops. This novel production method will invite new forms of broad collaboration through a partnership of multiple human and non-human agents; it also will provide a window into the cooperative research process, potentially helping demystify it and representing a new form of transparency in scholarly research.