The technology industry has spent years developing artificial intelligence. And we have already seen the first results, for example, Siri and virtual assistants. More recently we have seen other incredible achievements, such as Ms. Pac-Man’s perfect match and AlphaGo ‘s victory over the best Go player in the world.
However, what is clear is that humanity still has much to know in this almost unexplored domain. And one of the most fervent samples of it is out in a report of Facebook on a test between chatbots.
A test that ended in unexpected results
Facebook, being one of the main companies that have begun to invest in this area. It has an artificial intelligence research laboratory. And during a test where machine learning is used to teach its “dialogue agents” to negotiate, two chatbots. They initiated a conversation that led to the creation of their language.
At this point, the researchers had to stop the conversation. And change the model they used in that test because they could not follow the conversation they were having.
This event is by itself breathtaking. That is if artificial intelligence is sufficiently developed to be able to negotiate with people. This sample of evolution reveals only a part of what is to come in this area of industry. As we mentioned, humanity still has much to see in this area.
The project’s researchers conclude that “there is still a lot of potential for future work, particularly exploring other reasoning strategies. And improving the diversity of statements without departing from human language.”
Via The Atlantic