Seeking human touch in the digital age: UAE residents reveal desired traits in a human-like AI assistant

When interacting with human-like smart assistants, consumers in the UAE place high value on personalisation and accurate understanding, as revealed in a recent survey conducted by Researchscape International and commissioned by Yango. The survey aimed to discover the important features that UAE residents look for in an AI assistant.

The most significant human-like traits for respondents include the ability to recognise individual family members and respond accordingly (66%); accurately understand various accents, speech patterns, and intonations (63%); and provide chat support and offer assistance proactively (54%). Additionally, survey participants highlighted a sense of humour (43%) as well as the ability to understand whispered speech (34%) and respond in the same manner.

Having anticipated this need for personalisation, Yasmina's advanced capabilities enable the assistant to identify the user's gender and respond accordingly as well as provide age-appropriate content when interacting with children. Additionally, Yasmina excels at telling culturally relevant jokes that resonate with locals. When you ask the assistant to tell a joke, it responds with one from its carefully curated collection.

A majority of survey respondents speak Arabic at home, with Khaleeji (40%), Egyptian (32%), and Levantine (27%) being the most common dialects. 63% of the respondents stated the need for an assistant that speaks and understands Khaleeji Arabic. Yasmina excels at that: the AI assistant understands major Arabic dialects and speaks Khaleeji fluently. Yasmina can also seamlessly switch between Arabic and English during a conversation, providing responses in the appropriate language and maintaining natural and fluid dialogue.

Yasmina can maintain engaging, fun, and consistent dialogs that feel like talking to a real person. What makes Yasmina unique is its innovative natural language understanding, speech recognition, and speech synthesis technologies. Yasmina’s human-like conversation skills are powered by a proprietary large language model (LLM) that has been extensively trained on a rich and varied array of datasets.

“At the core of Yasmina lies a powerful large language model, which enables it to continuously learn and enhance its knowledge and skills. We have improved Yasmina's ability to generate natural language by exposing its LLM to diverse linguistic contexts that capture the intricacies of regional dialects, idioms, and cultural references”, said Samer Mohamad, Yasmina Regional Director for MENA, Yango.

Yasmina is currently available within Yango Play, an AI-powered entertainment super app combining video streaming, music, and mini-games in one place. However, the AI assistant will soon appear in more devices and apps, extending the list of its use cases and opening up new opportunities for engaging interactions.

*The findings are based on an online survey conducted between February 7th and 20th, 2024, among 500 participants from the UAE. The study was commissioned by Yango and carried out by Researchscape International.