Robot-produced languages as part of #newaesthetic?

As a follow up to my blogpost the other day about New Aesthetic as not-only-visual-but-also-something-else, I kept wondering about other possibilities. Overall, what I find interesting in NA is that algorithms produce new cultural forms… and that it’s not just about visual representations.

One of the cultural form that can produced by robots/algorithms for that matter is certainly language and communication. Researchers in Artificial Intelligence indeed work on that avenue for quite some time and it would be relevant to consider the “language” produced in this context by (ro)bots… and see whether it fits with the New Aesthetic meme.

A good starting point for this is Language Games for Autonomous Robots by Luc Steels:

s. A language game is a sequence of verbal interactions between two agents situated in a specific environment. Language games both integrate the various activities required for dialogue and ground unknown words or phrases in a specific context, which helps constrain possible meanings. Over the past five years, I have been working with a team to develop and test language games on progressively more sophisticated systems, from relatively simple camera-based systems to humanoid robots. The results of our work show that language games are a useful way to both understand and design human–robot interaction.

This is done through various experiments such as the “Talking Heads experiment” shown on the following picture. It’s hard to find a video of such work but this one might help to get a sense of what’s happening (it’s a guessing game though).

As shown by Steels in this paper:

During a three-month period, the agents played close to half a million language games and created a stable core vocabulary of 300 words (they generated thousands of words overall). Our experiment showed not only that the language game approach is useful for implementing grounded dialogues between one human and a robot, but also that the game might be useful as an explanatory model for how language originates

And this is exactly where you can find what can be part of New Aesthetic: the “language” that emerged from these robot interactions. One of my project as a graduate students in Cognitive Sciences was about that and I remember being fascinated by the work of researchers such as Frederic Kaplan or Bart de Boer. More specifically, the latter investigated emergent phonology, i.e. how iterations of imitation games in a population of agents can led to the emergence of sound systems emerge that look remarkably like human vowel systems. It’s not necessarily the fact that is used to model language evolution that interest me here. Instead, what is strikingly stunning is to see the results of such emergence and what it can produce.

Here’s the process:

The agents that are used in the computer simulation use vowels to “communicate” with each other. For this purpose, each agent has its own list of vowels. The lists of vowels for each agent are initially empty, and will be filled as the agents engages in interactions with other agents.
The experiments presented in this work are concerned with the emergence of a coherent and useful phonology in a population of initially empty agents. In order to investigate how this can happen, the agents engage in exchanges of sounds, so-called imitation games, the goal of which is to learn each other’s speech sounds. If necessary, speech sounds are invented, in order to get the communication started, and also in order to introduce more possible sounds in the population.

The result part of the paper is quite dense and describes the phonemes and vowels produced by the robots.

Why do I blog this? Of course, in this case, the goal is to model natural languages but it would be curious to play with the constraints to see how various sorts of language systems can emerge depending on the model parameters. And this is exactly where the result can be part of a cultural form that can belong to New Aesthetic.

On a different note, the problem, from the collection angle (that is quite important in the NA theme), is that it’s hard to find audio or video pieces that can help us to listen to these languages games. I’ll try to dig into that.