Who Benefits from AI? | Asia Summit 2026

Meta's Strategic Definition of Superintelligence

• Meta defines "superintelligence" as intelligence exceeding human capacity, specifically emphasizing "personal superintelligence" that is widely distributed to individuals rather than centralized.
• This philosophy aligns with a 6,500-word manifesto by Mark Zuckerberg advocating that the future of AI should be accessible to everyone.
• Meta has established the "Meta Superintelligence Lab" to focus on building these widely distributed capabilities.

Product Launches and Accessibility

• Meta released "Muse," an agentic AI product currently available in the U.S. and planned for Singapore, designed to execute complex tasks (e.g., scheduling, planning) rather than just answering prompts.
• Meta is launching "Llama Glasses" in partnership with Lisa from Black Pink; the devices cost approximately $250 but serve as a hardware entry point for personal superintelligence.
• The primary access point for Meta AI remains free via existing apps (Facebook, Instagram, WhatsApp) used by nearly 4 billion people globally.
• Muse is offered for free for most individual use cases, with premium pricing reserved for "power users."

Open Source vs. Closed Source Strategy

• Meta is doubling down on open-sourcing its foundation models, allowing developers, universities, and countries to download and customize them locally without API fees.
• This strategy contrasts with companies like OpenAI and Anthropic, whose business models rely on selling access to closed, frontier models.
• Meta argues that open-source models democratize technology by allowing users to control data, train on local data, and avoid dependency on U.S.-based servers.

Singapore's AI Policy and Economic Focus

• Singapore's government, led by Permanent Secretary Kai-Fong Cheung, is shifting focus from multinational corporations (MNCs) to supporting Small and Medium Enterprises (SMEs) and gig workers in AI adoption.
• The "SkillsFuture" initiative provides six months of premium AI subscriptions to all sign-ups to build basic literacy and confidence among the workforce.
• Singapore acknowledges the rising utility of "open-weight" models (e.g., Meta's Llama, Mistral, Chinese variants) due to their cost-efficiency and customizability for specific local use cases.
• Despite supporting open source, the Singapore government continues to use closed frontier models (e.g., OpenAI, Claude) for high-complexity tasks like government coding due to their superior sophistication.

Sovereignty and Digital Dependency

• While "sovereign AI" is often debated, Singapore views the concept as partially overstated; the priority is on the ability to switch vendors if necessary rather than total self-sufficiency.
• The adoption of open-weight models is seen as a critical risk-mitigation strategy, giving countries ownership over their data and the ability to post-train models on local data.
• Singapore is actively monitoring "track two" AI safety dialogues between the US and China, noting that Chinese labs are increasingly taking AI safety seriously with annual risk registers.

Regulatory and Safety Developments

• Joel Kaplan noted the recent voluntary White House accord on superintelligence as a significant step, where leading US labs committed to independent safety evaluations, internal assessment teams, and board-level oversight.
• Kaplan stated that while the accord is "morally binding," it will likely evolve into a liability regime in the US due to existing legal frameworks.
• A key diplomatic challenge remains securing similar commitments from Chinese frontier labs compared to the consensus among non-Chinese labs.

Economic Distribution and Productivity

• A recent Meta poll indicates that 80% of SMEs in the APAC region are currently using AI for back-office tasks, customer communication, and ordering, though most require assistance in optimizing usage.
• Kai-Fong Cheung noted that current revenue uplifts from AI are limited due to legacy organizational mindsets and the time required to reinvent existing business processes.
• Meta predicts a surge in productivity for the 200 million small businesses on its platforms if they can be effectively upskilled to use agentic AI tools.

Workforce Anxiety and Job Design

• There is acknowledged anxiety among workers regarding task displacement, necessitating a focus on job redesign rather than simple replacement.
• Kai-Fong Cheung warned that a steep acceleration in AI capabilities could cause economic dislocation, whereas a gradual progression allows for easier workforce adjustment.

Philosophical Outcomes: Human Flourishing

• Joel Kaplan envisions AI as a bridge to turn imagination into reality, freeing humans from mundane tasks to focus on fulfillment, invention, and creation.
• Kai-Fong Cheung warned against "AI psychosis" and the outsourcing of human thinking, emphasizing the need to double down on deep reflection, reading, and social capital building.
• Singapore is countering the "epidemic of loneliness" by promoting reading movements and social connections to ensure humans remain "more humans" despite technological integration.