Singapore’s National University Health System (NUHS) is leveraging advanced AI technologies to enhance efficiency and reduce administrative workloads in healthcare. Through the RUSSELL-GPT platform, which integrates large language models (LLMs) via Amazon Web Services (AWS) Bedrock, over a thousand clinicians now benefit from automated tasks such as drafting referrals and summarising patient data, reducing administrative time by 40%.
The NUHS team is working on event-driven Generative AI models that can perform tasks automatically when triggered by specific events, such as drafting discharge letters without needing any prompts. This approach aims to streamline processes further and reduce the administrative burden on healthcare staff.
Ensuring patient data security is a top priority for NUHS, with robust measures in place to keep data within Singapore and comply with local privacy laws. RUSSELL-GPT also includes features to mitigate the risks of AI hallucinations, with mandatory training for users on recognising and managing such occurrences.
Despite the promise of LLMs, NUHS acknowledges that these models are not a cure-all. Classical AI still plays a critical role in tasks like clustering information and providing predictive insights, underlining the need for a balanced use of it in healthcare.
SoftBank has abandoned its plan to develop an AI chip in partnership with Intel, according to a report by the Financial Times. The Japanese tech investor had intended to collaborate with Intel to challenge Nvidia, but the deal fell through after Intel failed to meet SoftBank’s requirements, as reported by sources familiar with the situation. SoftBank attributed the breakdown of negotiations to Intel’s inability to deliver on speed and production volume demands.
As a result, SoftBank has shifted its focus to discussions with Taiwan Semiconductor Manufacturing Co (TSMC), the world’s largest contract chipmaker. The report noted that the collapse of talks occurred before Intel’s significant cost-cutting measures, including massive layoffs in early August.
Why does this matter?
These events highlight the intensifying competition in the AI chip market, where companies like Nvidia currently dominate. SoftBank’s decision to abandon its partnership with Intel and shift focus to TSMC underscores the challenges Intel faces in keeping pace with AI-driven innovations. The move also signals potential shifts in global chip production dynamics, with TSMC further solidifying its role as a key player. Additionally, it reflects the broader implications of Intel’s internal struggles, such as meeting demand and cost-cutting, on its competitiveness in critical emerging technologies like AI.
The English Premier League is set to enhance offside decision-making with new technology from Genius Sports. Multiple iPhones, paired with advanced machine-learning models, will assist referees in making more accurate offside calls. Traditional Video Assistant Referee (VAR) systems have faced criticism for slow reviews and inconsistent decisions, leading to this shift.
Genius Sports developed ‘Semi-Assisted Offside Technology’ (SAOT) as part of its GeniusIQ system. Up to 28 iPhones will be placed around the pitch to generate 3D models of players, offering precise offside line determinations. Expensive 4K cameras will be replaced by iPhones, which capture between 7,000 and 10,000 data points per player.
Strategically positioned on custom rigs, iPhones will cover optimal areas of the pitch. Data collected will be processed by the GeniusIQ system, using predictive algorithms to assess player positions even when obscured. High framerate recording and local processing capabilities further enhance the system’s accuracy.
Genius Sports plans to fully implement the system in the Premier League by the end of the year. While the exact date remains unconfirmed, this marks a significant advancement in football technology, promising a more precise and consistent approach to offside rulings.
Google is intensifying its AI initiatives in India, with a focus on addressing language barriers and improving agricultural efficiency. Abhishek Bapna, Director of Product Management at Google DeepMind, emphasized the economic importance of breaking language barriers, particularly in areas like healthcare and banking. Google’s AI chatbot, Gemini, supports over 40 languages globally, including nine Indian languages, and aims to enhance language quality further.
In collaboration with the Indian Institute of Science, Google’s Project Vaani provides over 14,000 hours of speech data from 80 districts, empowering developers to create more efficient AI models for India’s multilingual environment. Additionally, the IndicGenBench benchmark helps fine-tune language models for Indian languages. These efforts are crucial to improving the accuracy and reach of AI in the country.
Google is also piloting its Agricultural Landscape Understanding (ALU) Research API in Telangana, designed to boost farm yields and enhance market access. The initiative aligns with Google’s broader goals of improving livelihoods and addressing climate change, offering granular data-driven insights at the farm field level.
These initiatives are expected to not only assist farmers but also attract end users like banks and insurance companies. Once the pilot program is completed, Google plans to scale the project to work with state governments across India.
Government agencies in Australia must disclose their use of AI within six months under a new policy effective from 1st September. The policy mandates that agencies prepare a transparency statement detailing their AI adoption and usage, which must be publicly accessible. Agencies must also designate a technology executive responsible for ensuring the policy’s implementation.
The transparency statements, updated annually or after significant changes, will include information on compliance, monitoring effectiveness, and measures to protect the public from potential AI-related harm. Although staff training on AI is strongly encouraged, it is not a mandatory requirement under the new policy.
The policy was developed in response to concerns about public trust, recognising that a lack of transparency and accountability in AI use could hinder its adoption. The government in Australia aims to position itself as a model of safe and responsible AI usage by integrating the new policy with existing frameworks and legislation.
Minister for Finance and the APS, Katy Gallagher, emphasised the importance of the policy in guiding agencies to use AI responsibly, ensuring Australians’ confidence in the government’s application of these technologies.
Two teams from Singapore have secured gold medals in the first-ever International Olympiad in AI, held in Bulgaria. The competition, which ran from 9 to 15 August 2024, gathered high school students from around the globe, with Singapore emerging as a top contender in the scientific round.
Students were selected through a rigorous test held on 28 April 2024 at Nanyang Technological University in Singapore. The successful candidates were then flown to Bulgaria to compete in various AI-related challenges, including natural language processing, machine learning, and computer vision.
Both Singaporean teams, Team Singapore 1 and Team Singapore 2, claimed gold medals, showcasing the country’s strong capabilities in AI. The teams were comprised of students Shah Mahir Hitesh, Kabir Jain, Ng Shi Yu, Wang Zerui, Glenn Wu, Wang Jiayu, Lam Eu Ler, and Chong Jia Shuo.
Besides Singapore, Poland also earned gold in the scientific round, while China and Vietnam received silver medals. Malaysia was among the teams awarded bronze, highlighting the competitive nature of the event.
WhatsApp is expanding its sticker options, offering users more ways to express themselves through its platform. Despite the availability of hundreds of emojis and sticker packs, many users may struggle to find the perfect expression for their emotions. To address this, WhatsApp has integrated AI and GIPHY, enhancing the experience.
Users can now access an extensive collection by tapping the sticker icon and searching with text or emojis. Additionally, WhatsApp allows users to create custom ones from their existing images. By simply opening a photo, the app automatically removes the background, leaving only the subject for further customisation.
They can be cropped, drawn upon, and decorated before being saved automatically in the sticker section. WhatsApp now lets users preview and reorganize packs by dragging them within the sticker tray, offering greater control over their collection.
The Hollywood actors’ union, SAG-AFTRA, has agreed with the online talent marketplace Narrativ, allowing actors to sell the rights to digitally replicate their voices using AI. The following deal addresses growing concerns among performers about the potential theft of their likenesses through AI, providing them with a way to earn income and retain control over how their voice replicas are used. Actors can set the price for their digital voice, ensuring it meets at least the union’s minimum pay standards, and advertisers must obtain consent for each use.
SAG-AFTRA has praised this agreement as a model for the ethical use of AI in advertising, emphasising the importance of safeguarding performers’ rights in the digital age. The issue of AI-driven voice replication has been a significant concern in Hollywood, highlighted by actress Scarlett Johansson’s accusations against OpenAI for the unauthorised use of her voice. That concern was also central to last year’s Hollywood strike and remains a key issue in ongoing labour disputes involving video game voice actors and motion-capture performers.
In response to the rise of AI-generated deepfakes and their potential misuse, the NO FAKES Act has been introduced in Congress, aiming to make unauthorised AI copying of a person’s voice and likeness illegal. The bill has gained support from major industry players, including SAG-AFTRA, Disney, and The Recording Academy, reflecting widespread concern over the implications of AI in entertainment and beyond.
Google has unveiled its latest lineup of Pixel smartphones, marking a significant shift towards deeper integration of artificial intelligence in its hardware. Unlike previous years, the company chose to announce the new models in the summer, positioning itself ahead of competitors as it races to incorporate AI technology across its products.
The new Pixel devices feature innovative AI-powered tools, including a unique function that allows users to search for information within screenshots. Additionally, Google’s chatbot, Gemini, can now be accessed as an overlay on other apps, offering assistance and generating content. The launch event, held at Alphabet’s Bay View campus, showcased these advancements, with Google’s senior vice president of devices and services, Rick Osterloh, emphasising the company’s commitment to practical AI applications.
With AI taking centre stage, Google’s event impressed industry experts, with some noting it as one of the most comprehensive presentations the company has ever held. This early release strategy comes as Google aims to stay ahead of its rivals, particularly Apple, which is expected to launch new AI features in its products later this year.
The Pixel 9 series includes several models, with the base version priced at $799, $100 more than its predecessor. The devices will start shipping in August, with the Pixel 9 Pro and Pixel 9 Pro Fold set for release in September, further highlighting Google’s push to lead in the AI-driven smartphone market.
Dutch copyright enforcement group BREIN has successfully taken down a large language dataset that trains AI models without proper permissions. The dataset contained information gathered from tens of thousands of books, news sites, and Dutch language subtitles from numerous films and TV series. BREIN’s Director, Bastiaan van Ramshorst, noted the difficulty in determining whether and how extensively AI companies had already used the dataset.
The removal comes as the EU prepares to enforce its AI Act, requiring companies to disclose the datasets used in training AI models. The person responsible for offering the Dutch dataset complied with a cease and desist order and removed it from the website where it was available.
Why does this matter?
The following action follows similar moves in other countries, such as Denmark, where a copyright protection group took down a large dataset called ‘Books3’ last year. BREIN did not disclose the individual’s identity behind the dataset, citing Dutch privacy regulations.
