Thinking about my dead self: Deep learning, and redesigning death in a chatbot
NewsNews
Share Facebook Share Twitter Share Google Plus Share Linkedin Link

In October 2016, I read a brief article in The Guardian about Eugenia Kuyda, an artificial intelligence expert and co-founder of Luda. The article was really more about a chatbot she had made in the wake of her friend’s death: a memorial bot for Roman Mazurenko. It was built in his likeness, with thousands of lines of his own writing from several years of text messages, donated to Kuyda by friends and family after his death. Deep learning and as many of Mazurenko’s digital fingerprints as could be collected, allows the friends he left behind to grieve in a special way. There is a beautiful article on The Verge about Kuyda and Mazurenko, and the bot that I’ve described.

 

Chatbots have been fuelling human curiosity

The articles in both The Guardian and The Verge made me think about my own digital attic of Me. I remember the day, many years ago, when I created my first email account. My mother was making gooseberry pie and I used this as inspiration for the ID. I spent hours every day, for several years on MSN Messenger talking to my friends. I spoke in chatrooms with strangers under various pseudonyms. I made a Bebo account and poured myself into it. I had an intimate back and forth every day for an entire year with a friend over Bebo messages. I joined the MMO gaming community and projected myself into these games. I got a mobile phone, I joined Facebook, Twitter, and I made a Gmail account. I installed, and uninstalled Kik at least six times. I used Viber, Whatsapp, and dating apps. My sincerest public efforts at writing can be found on Weebly, Medium, and my old college blog. I’ve been texting for twelve years: exactly half of my life, or roughly 4,380 days. All of this must mean something.

Chatbots have been fuelling human curiosity for decades; since ELIZA in the 60s, to the multitude of ordinary bots online today. And they are no longer a dream of sci-fi, or a niche interest of some programming whiz. My first encounter with a chatbot building platform was Snatchbot. I made a mental health chatbot, and I found the software so intuitive, that I ended up making more. The overwhelmingly persistent efforts to propel our understanding of artificial intelligence further and further into unfamiliar, and taboo territory is, to me, partially a reflection of our perception of knowledge as boundless, and our morbid curiosity with eternity and ‘what is’. In the film Chappie, the ability to capture and transfer consciousness allows humans to live outside of their bodies, to continue living in machines. The scarily possible future world of Black Mirror also toyed with the idea. Kuyda’s memorial bot is not the same thing — but it is toeing the same line.

It is a strange thought. If you were to gather all of the text messages and emails you have sent to friends, family and any other person in your address book, what kind of impression of you would this create? I feel that if all of this data were to be expressed visually, it would be one of those impressionist paintings that have a wild familiarity, like Kandinsky’s ‘Composition VII’. Or perhaps it is more likely to look like absolute empty space, white pastel on canvas — my dead digital self’s ego should not be inflated to such dizzying heights. Your digital fingerprint could be worth something in the future, as a lifelong log book that can be turned into a programme for others to interact with. Imagine, an impression of you being passed down on a USB stick through generations of your family: it feels like there’s a warm and fuzzy existential crisis wrapping itself around me.

Sign Up For Our Newsletter