Putting conversation-based research results on open source platforms
Amplification of user anxiety in’Science of Love’
It is an artificial intelligence (AI) chatbot (chat robot) created by Scatter Lab.
It turns out that ScatterLab has uploaded research results based on messenger conversations collected from its service on an open source platform. Points have been raised that the collected personal information may not have been sufficiently filtered.
Capture of the open source platform GitHub.
According to related industries on the 12th, it was confirmed that Scatter Lab uploaded a’artificial neural network model file’ learned from KakaoTalk conversation data on the open source platform’GitHub’. Artificial neural network is a machine learning technique that mimics human brain cells, and using this file, other people can create AI chatbots (chat robots) similar to “achieve”. This file was uploaded in September of last year, and it is believed that a large number of these files were shared due to the nature of the open source platform.
Scatterlab introduced this model through its blog last August. Scatter Lab said in a blog, “Our Ping-Pong team used a large amount of conversation data collected through the’Science of Love’ application as learning data.”
An official in the information technology (IT) industry said, “It is difficult to steal personal information from this file itself,” he said.
Meanwhile, ScatterLab acknowledged the use of user data, saying, “In the case of Iruda, learning was conducted based on the science text data of love,” through the statement. Scatter Lab, which operated the AI chatbot service Iruda, is known to have learned AI using 10 billion conversation data from users of its other service’Science of Love’.
Reporter Kim Nam-young [email protected]
Ⓒ Hankyung.com prohibits unauthorized reproduction and redistribution