Facebook shuts AI chatbots which got out of control after creating their own language of communication

Facebook shuts AI chatbots which got out of control after creating their own language of communication

Hi again, this time I am going to tell you a relatively old news which emerged in 2017 but even today raises concerns as to how we should use the AI. So, there was a research work going on, where while training certain chatbots for making them interact with humans more efficiently, the chatbots created their own language to communicate with each other from the scratch and without human input. The AI bots method of creating and communicating with the new language defied the provided codes.
If you are interested in going deep into this publicly released research work then you can access it here.

This started in the mid of the year 2017. Where Facebook announces that they have shut down a research work they were doing to make chatbots more efficient because they saw that the AI created its own language and were communicating with each other in short hand notations. Note that the short hand notations they used were not similar to what the humans use on the social platforms, rather the AI chatbots defined their own short hand notations that they were using to communicate to each other.

An Excerpt:

Here, is an excerpt from the conversation in the language created by AI:

BOB: "I can can I I everything else"

ALICE: "Balls have zero to me to me to me to me to me to me to me to me to"

The Objective:

Image credit: https://www.maxpixel.net/

The real objective for the researchers was to train the Artificial Agents were left to converse and negotiate with each other in an attempt to strengthen their conversational skills to chat in plain English text. But the AI agents(here chatbots) learned quite fast using trail and error method (given the amount of quality data they were being trained on), just exactly as how every human first learns to communicate in a language. The Technique they used was Reinforcement Learning where we train agents to try to maximize cumulative reward.

However, Facebook officially announced that they shut down this research work because they were not able to get the AI to work what they wanted it to (like making a bot that would sell insurance online without having to talk to human agent) and AI defied the sole purpose of the research.

But you can imagine what could happen if we make ALEXA ,CORTANA and SIRI talk to each other. They may end up creating their own language of communication not understandable by us(human beings).

What experts say:

Although this seems to be a huge leap for AI however two of the very known smart persons on earth have different opinions towards the AI:

• Elon Musk, who is always been on the forefront of warning people about the out of control AI and once told that AI is a rare case where we need to be proactive in regulation instead of reactive because if we're reactive in AI regulation then its too late (which was also supported by Professor Stephen Hawking who explained that as humans are limited by slow biological evolution, they could be superseded by AI.

• Mark Zuckerberg, who is frankly more in the AI business as Facebook is developing a lot of AI chatbots and some more devices, has a more optimistic view and considers that the point of fearing AI is irrelevant.

My Opinion:

So, in my opinion there can be two aspects of looking into this scenario:

First :

...that in the near future when we have to deal with efficiency for a task which requires AI machines to collaborate to work with each other, then we have to train the AI system to make its own language so that it can communicate with other AI systems just as we humans do today and this may become a necessary task for achieving the efficiency.

Second:

...that in the near future there will be AI machines which will be able to talk in a language not understandable by humans and we will not be able to figure out their motives of the AI robots working with us and there will be a constant fear just like when we step in another country where we are surrounded by people whose language is unknown to us and we are in a constant fear that what is the motive of this person and what are they talking about, are they talking about me? and so on... 

Conclusion:

Hence, I think that we are in a very important decade which is a turning point in the development of intelligent machines and we must be very conscious and precautious while dealing with the advancement of AI.

So, at last I would just say that its upto us as how we make our future, a brighter one or a dull one, and we should try to work upon it from the starting itself as everyone knows that :

Precaution is better than cure.

Let me know you thoughts in the comment section and feel free to ask any queries regarding technologies and I will respond with an answer.