Leah’s PhD: An Introduction

There have been tons of changes in my life recently.

I have once again moved countries, and have therefore spent the last few months trying to make some friends and figure out where I am (I’m still not entirely sure). In addition, I have somehow taken on the role of social organizer for all of these friends, and am constantly jumping from place to place, attending hockey games and house parties, and doing all of the other things that supposedly characterize the ‘student experience’.

This leads me to my most exciting news: I’m a student again!

I know. None of you saw that one coming.

Now that I’ve begun settling into my doctoral student status (I’ve already faced my first wave of ‘failure’), I can finally get back to blogging. Plus, my research has already provided me with some killer material for y’all. Get stoked.

My doctoral research focuses on algorithmic authorship and computer-generated texts. Remember SmarterChild? What about the train wreck that was Microsoft’s Tay? That kind of stuff.

SmarterChild, doin’ its thang.

I use SmarterChild and Tay as examples of algorithmic authorship because those types of bots (I’ll define ‘bot’ in a later post) are most people’s first, and most explicit, interactions with computer-generated texts.

SmarterChild pulled from a body of preprogrammed responses to interact with its users; it often rearranged users’ sentences to craft questions that encouraged users to elaborate upon what they had typed. SmarterChild wasn’t really that smart, but it offered hours of entertainment to preteens who has just been given permission to download AOL Instant Messenger and MSN Messenger.

Tay was significantly smarter. She(?) was programmed to learn from those who interacted with her via her Twitter account. Picking up words and phrases used across the Twittersphere, Tay would then produce her own tweets applying what she had learned. Occasionally, she would also tweet images, captioning them herself.

Unsurprisingly, Tay promptly began spewing racial slurs, references to drugs and sex, and calls for genocide. The dark web – that is, 4chan – had bombarded Tay with hateful content that forced Microsoft to suspend the account after only 16 hours. On a positive note, though, Microsoft’s machine learning algorithms were clearly working well.

A really tame Tay tweet. Click on the picture for some more tweets (which accompany a pretty good article).

I don’t actually work with bots for my doctoral research, but the programs that drive bots like Tay are similar to the ones I spend my time analyzing. Instead of considering computer-generated texts that come from two-way interaction, I’m considering non-interactive computer-generated texts that take the forms of ‘traditional’ texts like novels and news articles. My driving research question is:

What are the social and literary implications of algorithmic authorship?

It’s a broad question, and necessarily so. Few people have dedicated significant time to considering the human responses to computer-generated text output; so far, the emphasis has been on how to develop this technology. Computer scientists and programming enthusiasts have been goin’ to town, applying developments in artificial intelligence to create programs capable of producing computer-generated texts that seem indistinguishable from those authored by humans. Numerous companies have popped up, touting platforms that allow clients to produce personalized business reports and data-driven news articles in registers that are appropriate to each individual reader.

This is fantastic technology, and it fits well within in our current digital climate.

However, as the Tay example has proven, developers have not been necessary been thinking about how these technologies are actually being used, or how they will be used in the future. Sure, every technology is developed to meet a certain need. Yet, once those technologies are unleashed into the world, they are not always received in the intended ways. For example, a broom isn’t just used to clean floors: witches fly on them; Quidditch players waddle around with them between their legs as they chuck various balls to each other; curlers use them on the ice, to propel their rocks. The broom is an old technology that has been adapted to suit a variety of needs that couldn’t have been predicted by its inventor (unless Quidditch has actually existed since ancient times).

This is where my research fits in. I’m hoping to discern just how people receive computer-generated texts, and how computer-generated texts are changing our social and literary landscapes. I want to extend the conversation about algorithmic authorship to include a humanities perspective.

You may be wondering how this fits in with book history. Don’t worry – I’ve hardly abandoned my book history roots. If you’ve read my “WHY am I reading?” page, you’ll know that I have a broad view of what book history as a field should cover. For me, book history is all about knowledge transfer, and the sharing of non-fictional and fictional stories. My previous research has involved close consideration of the physical manifestations of knowledge transfer (e.g. manuscripts and printed codices), but I’ve since shifted my focus to a less physical manifestation: computer code.

There are a lot of reasons for my shift from manuscripts to computer-generated texts, but the main reason I’ve chosen to dedicate three years of my life to algorithmic authorship is because I believe that this one very practical application of book history to the real world. Book history provides the perfect foundation for a study of algorithmic authorship’s effects on social and literary landscapes because book history has been thinking about authors’ and readers’ and tangible artifacts’ effects on these landscapes since its conception as a field. This kind of study allows me to bungee jump out of academia’s ivory tower by looking to the future of the book in addition to the history of it.

During my studentship, I’ll be using this trusty old blog to keep everyone updated on my progress. This post has been a quick and dirty introduction (mostly to let you know that I’m still alive), but future posts will delve more into the nitty gritty.

For now, check out this presentation I did at my department’s first-year doctoral colloquium. Like all good PowerPoints, there’s very little text, but scrolling through the slides should nevertheless give you a feel for what I’ve done so far. Wanna talk about the slides? Get in touch!

Advertisements

One thought on “Leah’s PhD: An Introduction

Have something to say?

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s