Grab has launched its inaugural Artificial Intelligence Centre of Excellence (AI COE) located in Tower B, the newest wing in GrabHQ@one-north, with support from the Digital Industry Singapore.
The AI COE aims to accelerate AI-driven solutions that improve accessibility, boost productivity and growth, and contribute to smart nation building across Southeast Asia.
“With over 1,000 AI models powering our platform, we’ve seen firsthand how we can leverage AI to improve lives. Recent advancements in AI hold the promise to transform how we interact with technology, foster innovation and bridge digital divides,” said Suthen Thomas, CTO at Grab.
The AI COE is set to generate at least 50 high-value roles by 2025 across Product, Engineering, Data Science and Analytics. These positions will enable Singaporeans to work on transformative AI innovations that address Southeast Asia’s unique challenges, and can potentially be exported to other international markets.
Grab’s AI COE is committed to developing solutions that make digital services more accessible for underserved communities, including individuals with disabilities, the elderly and those less familiar with technology.
In collaboration with the Singapore Association of the Visually Handicapped (SAVH), the AI COE is developing a Voice Assistant feature to enable visually-impaired users to book Grab rides using voice commands, enhancing the accessibility of Grab’s services.
SAVH members played a pivotal role in testing and refining the feature to ensure it meets the needs of the community.
Built on OpenAI models, Grab has fine-tuned the speech-to-text model with 80,000 local voice samples and names of points of interest, boosting its recognition of Singaporean accents and building names from 46% to 89%.
To further enhance the model’s accuracy, Grab will launch a voice donation feature within the Grab app in June to enable Singapore users to contribute voice samples to help make Voice Assistant more intuitive and inclusive.
Voice Assistant is currently being piloted in Singapore, with plans to further expand its capabilities. Beyond supporting visually-impaired users, Grab is also exploring similar initiatives to simplify navigation for underserved communities, particularly the elderly.
Also, the AI COE is building a specialised foundational model, uniquely trained on Grab’s real-world data, to more accurately represent and interpret the behaviors of app users, driver and delivery-partners as well as merchant-partners.
This model will enable Grab to build AI solutions that offer hyper-personalised recommendations, improving user and partner experiences.
Additionally, the AI COE will also develop infrastructure and tools to support Grabbers on creative ideation and rapid experimentation of advanced AI technologies.
Grab’s in-house Large Language Model Kit, which contains pre-built code incorporating advanced AI models, enables Grabbers to set up a secure infrastructure to test AI-powered solutions in half a day, a reduction from 1.5 weeks previously.
