This browser is not actively supported anymore. For the best passle experience, we strongly recommend you upgrade your browser.
| 1 minute read

Sheepdogs, swarms, and semantics: better human-AI communication

How can AI and humans communicate better? According to a group of researchers, the answer lies in an Aboriginal language. 

In their article "JSwarm: A Jingulu-Inspired Human-AI-Teaming Language for Context-Aware Swarm Guidance" (Abbass et al., Front. Phys., 14 July 2022, Sec. Interdisciplinary Physics), the researchers propose a language ("JSwarm") for communication between humans and an AI agent based on Jingulu, a language spoken by the Jingili people of Australia's Northern Territory . 

Jingulu has a number of properties which make it ideal for bridging the gap between humans and AI. Firstly, it has a much freer word order than English, which requires sentences to be written using the "subject-verb-order" structure. (For example, "The attorney filed the patent application" is clear to an English speaker, whereas "filed the attorney the patent application" is not.) As a result, the meaning of a sentence in JSwarm is more robust to changes in word order. 

Secondly, verbs in Jingulu are constructed by modifying only three 'primary' verbs ("do"/"be", "go", and "come"). Therefore, JSwarm provides a computationally efficient way of communicating with an AI, but - being rooted in a natural language -  remains accessible for human users. 

The authors explain the new language using the case of a sheepdog cooperating with a human handler to herd sheep. However, the potential uses of JSwarm are far more extensive. For example, a real-life 'sheepdog' could be a chemical substance to guide a 'herd' of nano-robots treating cancer cells. Alternatively, the 'herd' could be a swarm of autonomous vehicles. Whatever the application, JSwarm remains a remarkable example of the ways in which traditional knowledge can spur technological developments. 

The focal point of this paper is human-AI teaming, especially within the context of distributed AI systems capable of synchronising actions to generate an outcome, or what we call AI-enabled swarm systems. In particular, our aim is to design a computationally efficient human-friendly language for human-AI teaming that is also appropriate for human-swarm interaction and swarm-guidance.


artificial intelligence, robotics, swarm robotics, traditional knowledge