Update 8 on the Fear-Disgust-Hate project
Background on “Fear” and “Disgust” as well as first drafts of short-text-responses to counter these feelings, are described in Update 1, Update 2, Update 3, Update 4, Update 5, Update 6 and Update 7.
The task I set for myself in the last post was:
… try to achieve this [hate-emotion] down-regulation using short-text-responses to disgust and fear (found in Update 3 and Update 6), hoping that the end result is the amelioration of hate.
…..then to test these short-text-responses on some profoundly hateful statements that seem to be triggered by disgust and hate, and see what happens.
Finding such a site and responding with some of these short-text-responses proved non-productive. Not that surprising, since my adventures this past year are those of a beginner and amateur with far to little time to do the leg work needed for building, even the foundations of, a project with ambitions on this scale. Thankfully, I believe I have come up with an alternative solution that makes me put this project in the “not a complete fail”-category.
Lucky for me, much smarter people (with appropriate professional skill sets) have done some impressive work on how to communicate efficiently in hate-contaminated settings. During this year I have discovered several, and especially Marshall Rosenberg stood out. He developed a whole theory of non-violent communication and wrote several books on the topic. Basically, he had described how to do exactly what I was setting out to do, many years ago. His work is not centered on web-based communication, but the principles are still valid.
Scrapping my half-assed text-responses for his non-violent communication theory as a basis for short-text replies, is probably a good solution.
Having AI use his theory to generate responses that are tailored to the original hate-containing text is probably an even better solution.
So without further ado, I pause this project at this stage with a public PartyRock App that lets you paste in any text, look for hateful language in the text and generate a response based on Rosenbergs non-violent communication theory.
You’ll find the app here: Non-Violent-Response-Generator
Below, I’ve pasted an example of how this looks for a random reddit-post.
The next phase could be to optimize the text responses and eventually to make an app with browsing capabilities able to directly scan web-sites and generate responses. But, that will have to be a project for later or for someone else to continue …
BIOpinionated 