An AI system that learns language autonomously develops a language structured in the same way as human language. There are parallels to the way that humans learn from previous generations in that AI models get better when they take advantage of the knowledge of ‘older relatives’ (previous AI versions).
This is shown by a study from Chalmers University of Technology and the University of Gothenburg, Sweden, which explores the mechanisms behind human language and provides important knowledge for the development of the AI of the future.
AI-based language models like ChatGPT are improving and they are becoming better at mimicking human language and are increasingly being used as a tool for producing text. But the same types of models can also teach us to better understand how human languages have evolved.
In the study, the researchers explored an AI system for evolving languages by using two different methods in a previously untested combination. On the one hand, reinforcement learning was used – where the right actions are rewarded and thus reinforced – and on the other hand, the researchers let the AI models learn from each other over generations.
This occurs because AI models learn in a way that is are structured in the same way as human language, and that language learning worked in a similar way to human language.
All languages strive to be efficient
According to an influential theory in cognitive science all human language is shaped by the need to be able to communicate effectively. At the same time, a balance is needed: as a tool, the language must be informative but also simple enough for us to learn. The more information that needs to be communicated, the more nuanced the language needs to be. A classic example is that languages in colder climates often have more words for snow and ice than languages in warmer climates.
The cognitive science theory of “efficient communication” measures the efficiency of language in a strictly mathematical way. According to the theory, all languages strive to be efficient. This means that, on the one hand, we want an informative language, on the other hand, a simple language, as this creates less effort and is easier to learn. According to efficiency theory, language strikes the perfect balance between these two parameters, and it can be different for different languages and cultures, depending on the needs involved.
To test the theory and investigate how language becomes efficient, the researchers created AI agents – different AI models – that played a communication game with each other. The AI agents were shown a colour and a list of symbols that initially had no meaning. As the agents interact, these symbols come to be associated with certain colours and used to communicate the colour to the other agent.
The reason why the researchers used colours is since there is so much data on how the colour spectrum is named in different languages, including data from isolated languages that have never been exposed to other languages. The categorisation of colours varies between languages, both in terms of the number of words and which part of the colour spectrum the words describe.
Rewards and generational exchange yielded results for language development
The experiments involved one AI agent communicating a colour via one of the symbols in the list, and the receiving AI agent would guess which colour the symbol corresponded to. Both agents received a common reward when they made progress in their communication. The closer they got to a common designation of the exact colour shade that the receiving agent got, the more points were awarded.
In the next step, new “generations” of AI agents were created, while the old AI agents were phased out. The new AI agents got to see the dialogue and the language that the previous generation had managed to develop. After that, the new AI agents got to play the same communication game with each other.
Provides knowledge about how language develops
The result was a system for naming colours that were similar to human colour languages, despite the fact that the AI agents had never come into contact with them. The results indicate that the human ability to communicate and learn from each other is crucial for how languages develop over time.
It is hoped that the results will contribute to new insights and ideas in language research, as well as research in AI and computer science.
The study “Cultural evolution via iterated learning and communication explains efficient color naming systems” has been published in the Journal of Language Evolution.
