Yesterday I used my new-found knowledge from the Computational Social Science course and did the following: I scraped all the content from my blog with a web scraper, and ran it through IBM Watson's Personality Insights. It purports to "gain insight into how and why people think, act, and feel the way they do. This service applies linguistic analytics and personality theory to infer attributes from a person's unstructured text. "
23,091 words were lifted from my blog and analysed. The summary of my personality is as follows:
"You are skeptical, shrewd and excitable.
You are philosophical: you are open to and intrigued by new ideas and love to explore them. You are authority-challenging: you prefer to challenge authority and traditional values to help bring about positive changes. And you are independent: you have a strong desire to have time to yourself.
You are motivated to seek out experiences that provide a strong feeling of discovery.
You are relatively unconcerned with both tradition and taking pleasure in life. You care more about making your own path than following what others have done. And you prefer activities with a purpose greater than just personal enjoyment."
Uh oh. It actually sounds quite like me, except that I think I take pleasure in life. Shit, the machine's on to me.
You are philosophical: you are open to and intrigued by new ideas and love to explore them. You are uncompromising: you think it is wrong to take advantage of others to get ahead. And you are empathetic: you feel what others feel and are compassionate towards them.
Your choices are driven by a desire for organization.
You are relatively unconcerned with both achieving success and taking pleasure in life. You make decisions with little regard for how they show off your talents. And you prefer activities with a purpose greater than just personal enjoyment.
Uh. Maybe this is Trump's speech writer we're talking about. Or maybe Trump is just a very, very good actor.
Today I watched this video, which talks about Natural Language Processing and the problems associated with it.
The researchers used four different services, by IBM Watson, Microsoft, Google, and Amazon, in a similar way as I did, putting in chunks of text, and generated sentiment analyses (whether a chunk of text is considered positive, negative, or neutral in tone) from the input text. Apparently the results across four services are wildly dissimilar, and all can be swayed in one direction or another by small things such as typos or grammatical errors. The machines also showed idiosyncratic tendencies, such as deciding that "my uncle made me feel irritated" is a more positive statement than "she made me feel irritated".
The problem is that these types of services are used a lot these days, whether in customer service chatbots or in a hiring process. The biased results may have unintended and severe consequences if we are not careful.
And also, nobody has to read my blog really. My labour of love, the condensation of my being, a peek into my endless wisdom... a bot can crawl through it and perform a "very strong analysis" and come up with a succinct summary of who I am. Or who my blog writer is. Or the persona that I am playing, as I am a very, very good actor.