Earlier this week, Google presented its latest project in fighting off internet trolls. This is called the Perspective API and it is part of the Conversation AI project. They are both developed by Jigsaw.
As the number of internet pages increases, so does, unfortunately, the number of trolls. The Internet can be quite a harsh place because of them. As such, many members of the industry have started taking measures.
And just recently, Google announced that it will be introducing a new tool. The reveal came this Thursday through the release of the Perspective API. This was developed by a Google subsidiary.
Called Jigsaw, it has a very clear purpose. The startup focuses on the way in which technology can help defend the international human rights. To this purpose, it established the Conversation AI project. Which also led to the appearance of the Perspective API.
CJ Adams went to offer some details. He is the Jigsaw product manager. And according to him, the Perspective API can be explained in quite a simple manner.
This is a tool that will basically analyze a comment. And based on this observation, it will return a score. Its scale will be set from 0 to 100. And depending on its value, users will determine its toxicity levels.
A more detailed explanation goes as follows. The Perspective API is based on artificial intelligence or AI. Before its release, it was loaded with a series of comments which are considered ‘toxic’ and which are seen as ‘trolling’. Due to its AI base, the tool should also continue learning. And as such, come to determine trolling on its own.
The Perspective API could come to be a very useful tool. It will be able to automatically rank a comment left in the respective page section. Or even on forums.
The API will then analyze it and determine the following. It will use the aforementioned scale system. Jigsaw calls it its ‘toxicity scale’. And it will try to establish if such a comment can be considered offensive. Or if it may actually make a person leave the conversation.
According to Jigsaw, the Perspective API will not work as an arbiter. It will not be establishing what a commenter can or shouldn’t say. Instead, it will simply be a tool. One that can be freely used by the online publishers.
For example, they can make it readily available to their readers. Which can use it to filter comments in accordance with their toxicity levels. If they wish it, they will only see non-toxic posts.
Online publishers could also use the API to return a feedback. This would announce the commenter if their comments can be considered toxic. It will not stop users from posting them. But it could make them rethink what they are writing. And the potential effects of their words.
Available online since February 23rd, the Perspective API is reportedly already being tried out. It is being utilized by Wikipedia. And also by The New York Times.
As it will start being used, this API will also become more exact. It should come to be able to understand context. And as such, tell the difference between a literal and figurative meaning.
Nonetheless, the Perspective API still has some work to do before it will reach this skill. But it will only advance once it starts being fed with information and comments. Through them, it may become able to understand the nuances and underlying context of a human conversation.
The Perspective API may also start being used in order to eliminate off-topic comments. Or ones that don’t have a general contribution to the conversation. It remains to be seen if or when Perspective will take off.
Image Source: Pixabay