I recently read an article about a journalist who created a chatbot to ensure that a conversational entity with elements of his dying father's personality would exist even after his father--terminally ill with cancer--passed away.

I am a software engineer experienced with creating chatbots and it has crossed my mind to create one for the parents of my best childhood friend who was killed 10 years ago in a car accident.

I speak to the parents a few times a year and we have a great relationship. Having been their son's best friend when we were growing up, I know his mannerisms and expressions almost as well as they do so I am confident that I could do a good job of training the chatbot.

That said, the parents have found their peace with my friend's passing and I don't want to do anything to jeopardize that. In any case, I would obviously discuss with them before creating such a chatbot. This leads to my ultimate question:

Would creating a chatbot based on a deceased loved one help or harm those of us left behind?

I really don't know the right answer to this. In my most generous thoughts of how this turns out, it could give his parents, siblings, and future family members a way to remember/meet his fun, goofy personality, but I could also see this opening old wounds.

  • 1
    $\begingroup$ If you have Netflix I would like to point out the serie Black Mirror, especially episode "Be Right Back " which is about exactly your question. The show is a drama which holds a mirror in front of us, showing us the negative and scary consequences that technology has on humanity and society. Very interesting and definitely recommend. $\endgroup$ – Robin Kramer Sep 2 '17 at 6:18
  • $\begingroup$ And regarding the answer, perhaps there is some study on attachment to memento or the like, with conversations being an odd variant of that. $\endgroup$ – Robin Kramer Sep 2 '17 at 6:48
  • 1
    $\begingroup$ Certainly this depends on the person, use, etc, in addition to the somewhat arbitrary definition of "healthy"? $\endgroup$ – mflo-ByeSE Sep 2 '17 at 19:35
  • $\begingroup$ For my 2 cents...i can't see it as healthy. We must bury our dead as all humans before us have. $\endgroup$ – JMJ Sep 3 '17 at 1:06
  • $\begingroup$ Since this question is not framed in the cognitive sciences, answers (as can be seen) are primarily opinion-based. I'm afraid this question is not a good fit for this site, but please comment if others disagree. $\endgroup$ – Steven Jeuris Sep 7 '17 at 9:54

In my opinion, it depends on whether they would like the AI to synthesize novel thoughts/opinions or to merely contextually quote a history of that individual's textual recordings.

Making a value judgement here, I would say than an AI saying "I forgive you." (from beyond the grave) is not healthy.

But if the chatbot acted as a way of indexing into the individual's actual history, then all potential errors become ones of context (and not content).
(This level of reproduction may not be totally necessary depending on the users' sensitivity to cognitive dissonance, but I would personally feel slimy if I tried to pass off a markov chain of scripts and data as my friend.)

If you're open to alternatives to chatbots, I would suggest that you frame the project as a tribute to your friend's life; even a simple collection of stories from your time together could have a powerful (healthy) impact on the grieving parents.


Not the answer you're looking for? Browse other questions tagged or ask your own question.