While developing negotiating chatbot agents, Facebook researchers found that the bots spontaneously developed their own non-human language as they improved their techniques, highlighting how little we still know about how artificial intelligences learn.
THE FUTURE OF LANGUAGE
Using machine learning algorithms, dialog agents were left to converse freely in an attempt to strengthen their conversational skills. Over time, the bots began to deviate from the scripted norms and in doing so, started communicating in an entirely new language — one they created without human input. In a language sense, this one is mostly gibberish. But it’s interesting that AI, if given the opportunity, begins to deviate from the script to create something new — even if it looks like this:
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
Bob: you i everything else . . . . . . . . . . . . . .
Alice: balls have a ball to me to me to me to me to me to me to me
Bob: i i can i i i everything else . . . . . . . . . . . . . .
Alice: balls have a ball to me to me to me to me to me to me to me
Bob: i . . . . . . . . . . . . . . . . . . .
Alice: balls have zero to me to me to me to me to me to me to me to me to
Bob: you i i i i i everything else . . . . . . . . . . . . . .
Alice: balls have 0 to me to me to me to me to me to me to me to me to
Bob: you i 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
In an attempt to better converse with humans, chatbots took it a step further and got better at communicating without them — in their own sort of way.
A recent Facebook report on the way chatbots converse with each other has given the world a glimpse into the future of language.
In the report, researchers from the Facebook Artificial Intelligence Research lab (FAIR) describe training their chatbot “dialog agents” to negotiate using machine learning. The chatbots were eager and successful dealmaking pupils, but the researchers eventually realized they needed to tweak their model because the bots were creating their own negotiation language, diverting from human languages.
To put it another way, when they used a model that allowed the chatbots to converse freely, using machine learning to incrementally improve their conversational negotiation strategies as they chatted, the bots eventually created and used their own non-human language.
The unique, spontaneous development of a non-human language was probably the most baffling and thrilling development for the researchers, but it wasn’t the only one. The chatbots also proved to be smart about negotiating and used advanced strategies to improve their outcomes. For example, a bot might pretend to be interested in something that had no value to it in order to be able to “sacrifice” that thing later as part of a compromise.
Although Facebook’s bargain-hunting bots aren’t a sign of an imminent singularity — or anything even approaching that level of sophistication — they are significant, in part because they prove once again that an important realm we once assumed was solely the domain of humans, language, is definitely a shared space. This discovery also highlights how much we still don’t know about the ways that artificial intelligences (AIs) think and learn, even when we create them and model them after ourselves.
