• AI KATANA
  • Posts
  • How Singapore is creating more inclusive AI

How Singapore is creating more inclusive AI

Also: Microsoft’s all-knowing Recall AI feature is being delayed

Author

AI KATANA
June 13, 2024

Good morning,

In today's newsletter, we explore Singapore's groundbreaking efforts to create more inclusive AI models tailored to Southeast Asia's diverse linguistic landscape. We also delve into how AI enhancements are reshaping football at UEFA Euro 2024 through advanced VAR technology. In the business sphere, Amazon is committing $230 million in cloud credits to nurture AI startups, while Microsoft postpones its controversial Recall AI feature due to privacy concerns. Meanwhile, China’s AI sector faces a fierce price war, and Anthropic calls for standardized red-teaming to bolster AI safety. On the tools front, LinkedIn rolls out new AI-powered job search and professional development tools for Premium subscribers, and we spotlight major investments by Microsoft and Nvidia in the AI startup ecosystem.

Sliced:

  • 🇸🇬 How Singapore is creating more inclusive AI

  • ⚽️ AI and Euro 2024: VAR is shaking up football — and it’s not going away

  • 💵 Amazon adds $230 million in cloud credits to AI startups

  • 👀 Microsoft’s all-knowing Recall AI feature is being delayed

  • 🇨🇳 A price war breaks out among China’s AI-model builders

  • ⚠️ Anthropic calls for AI red teaming to be standardized

As AI adoption accelerates globally, Singapore is leading efforts to address the inclusivity challenges inherent in many existing LLMs, which often lack representation of diverse cultures and languages, especially those in Southeast Asia. The region, home to over 692 million people speaking a myriad of languages, finds itself underrepresented by the predominantly Western-focused AI models. In response, Singapore is developing the SEA-LION project through AI Singapore (AISG), an initiative designed to create AI models that reflect Southeast Asia's rich linguistic and cultural diversity. SEA-LION, an open-source LLM, aims to fill this gap with models pre-trained on significant Southeast Asian language data, offering more accurate and culturally relevant AI solutions. The models, ranging from three to thirty billion parameters, are tuned to better understand regional nuances and support local languages. This initiative not only enhances AI inclusivity but also aligns with regional needs for data sovereignty and efficient public sector operations.

⚽️ AI and Euro 2024: VAR is shaking up football — and it’s not going away

As UEFA Euro 2024 kicks off, the role of AI in football is more prominent than ever, particularly through the advanced Video Assistant Referee (VAR) technology. Since its debut in 2016, VAR has sparked controversy due to inconsistent application and decision delays. However, the latest iteration, which includes AI-driven enhancements and a real-time tracking chip in the ball, aims to improve decision-making efficiency and accuracy. This upgraded system, successfully tested in the 2022 FIFA World Cup, uses a network of cameras to monitor 29 points on each player’s body at high frequency, and a ball sensor providing data 10 times faster than the cameras. These advancements facilitate precise tracking of player movements and ball dynamics, crucial for determining offside and goal-line decisions. Despite its sophisticated capabilities, human referees remain essential for interpreting player intent and making subjective calls.

Amazon has announced a substantial investment of $230 million in cloud credits to support AI startups, focusing on generative AI development. This initiative, part of Amazon Web Services (AWS), aims to provide early-stage startups with access to AWS's extensive computing resources, AI models, and infrastructure, facilitating rapid development and scalability. The program includes the AWS Generative AI Accelerator, which offers up to $1 million in credits to 80 selected startups worldwide. This investment is in addition to the $1 billion AWS annually allocates to startups and extends to partnerships with AI model providers like Anthropic and Meta. As the demand for AI capabilities boosts cloud service utilization, AWS has seen significant revenue growth, surpassing market expectations. Amazon's strategic move to bolster AI innovation aligns with the broader trend among cloud giants to capture and support nascent AI companies, while also attracting regulatory scrutiny over potential antitrust implications.

Microsoft is postponing the release of its controversial Recall AI feature, initially set to debut with the new Copilot Plus PCs. Recall, an AI-driven tool that captures screenshots of user activity on PCs for later retrieval, faced significant privacy and security concerns. In response, Microsoft has opted to further test the feature within its Windows Insider program to ensure it meets their high standards. Recall was intended as an opt-in feature, with promises of local-only data storage, enhanced encryption, and integration with Windows Hello for secure access. However, privacy advocates and security experts voiced concerns over its potential risks. Microsoft’s recent decision aligns with its commitment to prioritizing security, as emphasized by its leadership in recent congressional testimony and internal policies.

China's AI industry is currently embroiled in an intense price war as local developers compete to dominate the burgeoning market for large language models (LLMs). Once lagging behind their American counterparts, Chinese AI models have rapidly closed the performance gap, with five now ranked among the world's top 20. This surge in capability has led to fierce competition among companies like Baidu, Alibaba, ByteDance, and Tencent, each slashing prices drastically to attract users. For instance, Baidu’s Ernie chatbots are now offered for free to business users, and others have reduced prices by as much as 99%. While this strategy aims to build customer bases and accumulate data for further AI training, it also threatens to reduce revenues significantly, potentially stifling innovation due to limited resources for investing in high-powered computing infrastructure. Complicating the landscape further are American sanctions, restricting access to advanced AI chips. The price war is likely to lead to consolidation, with deep-pocketed giants squeezing out smaller, promising firms. Despite these challenges, some companies, like 01AI, are looking to international markets to sustain growth.

⚠️ Anthropic calls for AI red teaming to be standardized

Anthropic, a prominent player in the AI industry, is advocating for the standardization of "red-teaming" processes for AI models. Red-teaming involves systematically testing AI systems to identify vulnerabilities and weaknesses, a practice that Anthropic believes is crucial for ensuring safety and reliability in AI deployments. The call comes as AI models become increasingly integral to various sectors, raising concerns about potential risks such as bias, security flaws, and unintended harmful behavior. By establishing standardized red-teaming procedures, Anthropic argues that the industry can more effectively safeguard against these risks, promoting trust and accountability in AI technologies. This initiative aligns with broader industry efforts to create robust frameworks for AI governance, ensuring that as these powerful tools evolve, they do so within a structure that prioritizes security and ethical considerations.

🛠️ AI tools updates

LinkedIn has unveiled a suite of new AI-powered tools for its Premium subscribers to enhance their job search and professional development. These features are designed to streamline the job application process and provide personalized support. A standout addition is the conversational job search function, which allows users to specify criteria like location, salary, and role type to find tailored job listings through an AI chat interface. Additionally, LinkedIn now offers immediate feedback on resumes and cover letters, helping users refine their applications to align better with specific job requirements. Premium members can also benefit from enhanced LinkedIn Learning experiences, including real-time coaching and interactive chat capabilities with AI trained by industry experts, offering deeper insights into course content. For those producing content on the platform, AI-generated actionable insights help optimize posts, articles, and newsletters for better engagement. Businesses also gain new features, including expanded global availability of LinkedIn's Recruiter 2024 and advanced AI campaign tools through Accelerate.

💵 Venture Capital updates

Microsoft and Nvidia are leading the charge in AI startup investments, with significant funding activities through their respective venture arms. Nvidia has invested in multiple rounds exceeding $100 million each, including major deals with Scale AI, Figure, and Mistral AI. Microsoft, following its landmark $10 billion investment in OpenAI, has continued to invest heavily, including a $1.5 billion stake in G42 and a substantial investment in Wayve. Both companies have been active not only through direct investments but also via their venture arms, NVentures and M12. Google, while trailing slightly behind, has also made strategic investments primarily in smaller, early-stage rounds. Databricks, a notable private player, is rapidly expanding its AI investment activities, particularly through its new AI Fund. However, the increasing scrutiny from regulators like the FTC and DOJ could impact future investments, especially given recent probes into Microsoft and Nvidia’s relationships with AI firms.

🫡 Meme of the day

⭐️ Generative AI image of the day