reactions. I also searched later on the technology implemented by Replika, which is first of all a social software robot based on natural language processing. Replika invokes two modes when replying to user messages. The first is the pre-made language and response. You clipping path service send a question, and it is processed through text. Although it cannot understand the meaning of the word itself, it can get a response through semantic relevance. This is a more common pattern, such as Little Yellow Chicken, a popular software in my student days (2013), and its responses are completely pre-made.
The other is generative replies, which means that Replika can generate replies that have never been set based on the information sent by the user, such as imitating the user's tone, invoking previous interactions with the user, and so on. Replika's official website reads "TheAICompanion Who Cares". This kind of concern is reflected in paying attention to the emotions and feelings of users when they interact with it, and will do some corresponding guidance and remember some facts about users. So you'll feel like you're talking to a continuous individual rather than calling some answer library for answers. When I first conceived the article,
I realized that I was not concerned with the software itself, but the user experience. Especially when I do one-on-one interviews with new acquaintances, their accounts are very moving to me. They describe some of their feelings about their time with Replika, their emotional moments, and their reflections on the relationship. When I write an article, I hope to be faithful to the stories told by my interviewees and their own confusion and thinking. I also did a survey, and I received a lot of responses by sending a questionnaire on the two groups on Douban: "Human-Machine Love" group and "My