A groundbreaking AI system developed by researchers at Ben-Gurion University of the Negev has demonstrated its ability to identify violations of social norms
The system uses GPT-3, zero-shot text classification, and automatic rule discovery to categorise social emotions into ten main types. It accurately evaluates written situations as positive or negative based on these categories.
Fundamental breakthroughs in the AI system
The AI system operates based on ten categories of social emotions, enabling it to detect social norm violations precisely. This innovative approach significantly advances cross-cultural language understanding and situational awareness.
Advancing cross-cultural language understanding
Funded by DARPA, the project is among the first to focus on automatically identifying social norm violations. While many social norms exist worldwide, the research in AI found they can be boiled down to general categories.
By employing advanced technologies such as GPT-3 and zero-shot text classification, the AI system can analyse short texts and efficiently identify whether they involve violations of specific social norms.
Impactful applications and validation in AI
The research holds particular importance in improving cross-cultural interactions and promoting effective communication.
The DARPA-funded Computational Cultural Understanding (CCU) program aims to enhance situational awareness for Department of Defense operators.
The AI system is crucial in fostering better international relations by preventing cross-cultural miscommunication, which can lead to diplomatic conflicts or even war.
Artificial intelligence and the ten social emotions
Under the guidance of Prof. Yair Neuman, head of The Functor Lab in the Department of Cognitive and Brain Sciences at BGU, the system was trained to identify ten social emotions:
- competence
- politeness
- trust
- discipline
- caring
- agreeableness
- success
- conformity
- decency
- loyalty
With these classifiers in place, the AI system effectively evaluated written situations and accurately determined their positive or negative nature.
Promising outcomes and prospects in AI
The results of this initial study have shown great promise, offering compelling evidence that the approach can be further expanded to encompass additional social norms.
The successful validation of the AI system on large datasets of short texts has paved the way for its potential application in various fields, including diplomacy, conflict resolution, and cross-cultural communication.
Automation: A leap towards improved social understanding
In conclusion, the AI system’s ability to identify social norm violations represents a significant step forward in cross-cultural language understanding and situational awareness.
With the potential to enhance interactions between individuals and nations, this pioneering research opens up new possibilities for promoting mutual respect and effective communication in a diverse world. The prospect of a more harmonious and interconnected global society looms as technology evolves.
