A Facebook chatbot speaking its own language should not be a cause for concern. If anything, it should be celebrated.
As ever, critics have been quick to jump on the “artificial intelligence is the end of humanity” bandwagon following Facebook’s decision to shut down two chatbots which had started communicating in an unfamiliar language.
Here is a snippet of the Facebook chatbot conversation:
- Bob: i 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
Some writers have warned that this is the start of artificial intelligence (AI) “going rogue” and “turning against its creators.” While the conversation content may appear rather creepy, the reality is quite different.
Facebook chatbot: Did Facebook “panic” and shut down its chatbots?
Yes, Facebook did discontinue the two chatbots but not over concerns about the content of their conversations.
As explained previously, the agents were designed to observe whether or not bots with differing goals could engage in negotiations with other bots or people, arriving at common decisions or outcomes. Facebook simply shut down the conversation for this reason – the language the bots were using was not productive in order to engage with humans.
As explained by researcher Dhruv Batra, there was nothing to be gained for the purpose of their project in having bots which could not be understood by humans.
In a Facebook post, Batra stated that agents like the Facebook chatbot will find ways to maximize reward and changing the parameters of an experiment is not equivalent to shutting down AI. In essence, robots will always find ways to achieve the goals set by humans.
Is the prospect of AI speaking a language of its own a bad thing?
Not necessarily. The prospect of building software that can communicate with each other more efficiently should be an enticing one.
In principle, the future surrounding this robotic language is incredibly short-sighted. At the heart of the technological world is the desire to maximize efficiency in order to help push the boundaries of innovation. If AI-powered machines are able to perform a task quicker without “threatening human life” then why should they be stopped?
We have seen multiple examples of computers being able to form their own languages – as outlined here. Therefore, it is far from reasonable to envisage a scenario where, instead of investing significant time and money on developing APIs, different software and apps are able to communicate with each other to provide a more seamless experience.
There are other, more important headaches surrounding AI to consider; robots spouting harmless gibberish is not one of them.
Companies around the world including AMP NZ utilize the InbentaBot to maintain a personal service for their customers while reducing support tickets.
Interested in finding out more? Our team of experts is at your service to design a custom proposal for you.
Check out our similar articles
- 8 Must-Have Chatbot Features to Improve Customer Experience
- Why Self-Serve Support Is Only as Good as Content
- The Most Popular NLP Use Cases