The Challenge
Create a series of posts on a microblog with the assistance of predictive text.
To be a true microblog I created my own Twitter account for this task and recorded the process of using predictive text following the provided starting words:
- Education is not about…
- My idea of technology is…
- Every time I think about our future…
- As a society, we are…
- This is not my idea of…
Predictive text in action:
Reflection
In a similar fashion to McRaney’s comments on predictive text suggesting a doctor to be male, I too noticed a stereotype where the word education seemed to suggest someone that was female with the recommendations of adding “she” and “her” to the text . McRaney mentions that predictive text is not an artificial intelligence but an algorithm based on a history of texts and as a result I believe this is where such stereotypes on job genders is coming from. Oddly enough the predictive text recommended “Blackberry” more than once even though it had no involvement with the starting text aside from the mention of technology.
The final posts that were created did not sound like me and they also did not make much sense as it felt like I was building a puzzle with pieces that were a mixture of several different puzzle sets. I think the algorithm may work in a sense of suggesting text that the user may want and I have noticed that it does start to suggest relevant words on topics I do often text about. For example, it suggested “green” a few times during my experiment, a topic that I often discuss in relation to environmental topics and sustainability.
I have been using social media apps, such as Twitter for a while and I appreciate the format because it allows me to consume information in small bites. This is helpful when I don’t have lots of time to read full articles or journals. The posts that the predictive text created did not remind me of any other text but it did get me thinking about social media bots, programmed accounts that will share, retweet, and post about topics that it was designed to share. For example @DearAssistant is a Twitter bot that will attempt to answer other users that ask questions in a similar fashion to digital assistants like Siri.
In the end, though predictive text can be helpful, it often seems to get in the way, much like the autocorrect feature of many digital devices.
Reference
McRaney, D. (n.d.). Machine Bias (rebroadcast). In You Are Not so Smart. Retrieved fromhttps://soundcloud.com/youarenotsosmart/140-machine-bias-rebroadcast