Filtered by: Department of Computer Science
Professor Anthony Tung from the Department of Computer Science at NUS Computing was featured in a Lianhe Zaobao report examining the growing spread of AI-generated fake videos and their potential impact on public trust and political discourse.
The report highlighted how advances in generative AI have lowered the cost and effort required to produce highly convincing misinformation, including fabricated videos that combine synthetic scripts, voices, and visuals. Such content, often designed to provoke emotional reactions, can spread rapidly on social media platforms and distort public understanding.
Professor Tung noted that heightened attention around recent political developments naturally drives engagement with such content, particularly among audiences with anti-establishment leanings. He added that Singapore’s high educational attainment means most citizens are able to exercise discernment when encountering sensational or implausible claims.
Beyond monetising online traffic, Professor cautioned that some videos may conceal deceptive schemes, using fabricated news narratives to promote products or services in ways that are difficult to guard against. He observed that regulating such content remains challenging, and suggested that one of the most effective responses may be to “counter AI with AI” — drawing parallels to how spam emails are filtered rather than prevented.
The report also noted that Singapore’s relatively high media literacy provides some resilience, but warned that misinformation may also be used to mask scams or malicious links. Overall, the report emphasised the importance of public awareness, media literacy, and the development of tools to help detect AI-generated misinformation, as synthetic media becomes increasingly sophisticated.
Associate Professor Suranga Nanayakkara from NUS Computing was featured in The Straits Times for his work on an AI-powered wearable headset developed by an NUS research team led by him, aimed at supporting independent navigation and daily activities for persons who are visually impaired.
The article highlighted the experience of Madam Teresa Ng, who shared how the headset could help her commute, shop, and navigate public spaces with greater confidence and independence, reducing the need to rely on assistance from others.
"It could tell me the price, size, colour, and even describe the kin of occasion each outfit is suitable for – very useful. Sometimes when I go grocery shopping, I have to ask the promoters for help and later feel obligated to buy from them. But with AiSee, I won't have to trouble anyone."
The wearable headset, AiSee, uses AI to interpret the surrounding environment and provide real-time audio guidance, helping users navigate public transport, avoid obstacles, and move safely through complex everyday settings. The system is designed to improve mobility and autonomy while preserving users' dignity.
Associate Professor Nanayakkara explained that the project was developed with a strong human-centred focus, emphasising usability, real-world impact and close collaboration with visually impaired users to ensure the technology addresses practical needs rather than laboratory assumptions.
Funded by the Enabling Lives Initiative (ELI) Grant in November 2025, the team is enhancing the AiSee software with onboard object detection, navigation, expert guidance, scene sonification, and task-based functionalities, including pilot testing at the Singapore Botanic Gardens to guide users on curated tours. The ELI Grant, funded by Tote Board and managed by SG Enable, is a strategic fund that catalyses and supports innovations to improve the independence and inclusion of persons with disabilities.
This coverage highlighted the potential of assistive AI to enhance inclusion and accessibility, while recognising the importance of careful design, testing and deployment in complex real-world environments.
Associate Professor Harold Soh from NUS Computing provided expert perspectives in CNA's report that examined whether automation and robotics can help address manpower shortages in Singapore's food centres.
Commenting on trials such as a self-cleaning table prototype at NTU, Associate Professor Soh described food courts as a particularly challenging environment for robots.
"Effective cleaning in such a space requires both physical intelligence (dexterous hands, balance) and social intellligence (knowing when to move, stop, and how not to annoy or injure customers), he said. "Robots today struggle with both."
Associate Professor Soh added that while robots can perform impressive feats in controlled settings, they continue to struggle with everyday tasks such as handling irregular food waste, managing spills, and operating safely in dynamic, crowded public environments. Integrating perception, prediction, planning and human-aware interaction remains an unsolved challenge outside the labs.
The report highlighted that while automation may assist with routine tasks, human cleaners and supervision remains essential, emphasising the technology alone is not a simple solution to labour constraints in food centres.
On 8world (Channel 8 News), Professor Anthony Tung from NUS Computing provided expert insights in an SG60 segment that examined how technology has transformed Singapore's society over the past six decades, from early office automation to the rapid rise of artificial intelligence (AI).
Representing academia, Professor Tung commented on the future direction of AI, emphasising the potential of AI systems to unlock collective potential by connecting people across disciplines, enabling lifelong learning beyond a single field of expertise.
The segment also noted that while AI adoption is accelerating across industries, it is not a panacea, highlighting the need for continued exploration and thoughtful application to meaningfully improve lives.
Professor Ng Hwee Tou from NUS Computing shared his expert perspectives in CGTN's (China Global Television Network) Asia Today segment on Singapore's AI pivot, which focuses on the country's efforts to upgrade AI models for local languages using Alibaba's Qwen3.
He explained why effective translation in multilingual contexts must go beyond word-for-word conversion to capture meaning, cultural references, and local expressions. His commentary was provided in the context of Qwen3's ability to support 119 languages and dialects, highlighting how multilingual AI models are designed to better handle linguistic nuance compared to many Western systems.
