The New Digital Divide: AI’s Haves and Have-Nots

The burgeoning digital divide, exacerbated by the rapid ascent of artificial intelligence, presents a formidable challenge to global equity. As Haslinda Amin, Anchor & Editor-at-Large at Bloomberg Television, starkly articulated, "AI is the tech revolution of our time," promising trillions in economic value, yet leaving over a hundred countries "lagging way behind." Even within advanced economies, significant disparities exist, with 30% of US internet users experiencing speeds too slow even for basic video conferencing, let alone sophisticated AI applications. This foundational inequity risks creating an entrenched class of AI "haves" and "have-nots," deepening global imbalances.

Lila Ibrahim, Chief Operating Officer of Google DeepMind, and Josephine Teo, Minister for Digital Development & Information, Republic of Singapore, engaged with Amin at the 2025 Bloomberg New Economy Forum in Singapore. Their discussion centered on navigating this critical juncture, exploring strategies to bridge the chasm of unequal AI access and foster a more inclusive technological future.

A core insight emerging from the conversation is the imperative for AI development to be inherently global and collaborative, rather than an afterthought. Ibrahim stressed that "AI is not a technology where we have national borders. You launch a model and it's available worldwide." This necessitates a proactive approach to ensure diverse representation from the outset. Google DeepMind’s partnership with AI Singapore, focusing on fine-tuning open-source models with Southeast Asian languages, exemplifies this commitment. Such initiatives prioritize data inclusivity and multimodal interaction, allowing users to engage with AI through speech, text, or images, thereby lowering the barrier to entry globally.

Beyond mere access, the discourse underscored the critical role of developing comprehensive capabilities. Minister Teo acknowledged the undeniable importance of AI infrastructure, particularly compute capacity, in determining who gains "full access" to AI. However, she posited that nations are not without agency to reshape their readiness.

"The single most important area of intervention… is really in developing capabilities," Teo stated, emphasizing a holistic approach. This encompasses fostering capabilities across the workforce, enterprises, academia, and within government itself, spanning entrepreneurial drive to advanced engineering. These are the fundamental building blocks that allow a country to truly harness AI's potential.

The conversation further explored the tension between open and closed AI models. While open-source models promise wider accessibility and faster innovation, concerns around data security and cyber breaches persist. Ibrahim highlighted Google DeepMind's commitment to releasing advanced, thoroughly tested models like Gemini 3, alongside open-source initiatives like AlphaFold, which democratizes access to protein prediction for millions of researchers worldwide. Teo championed open models for widespread adoption, arguing that without them, achieving broad societal benefit from AI would be problematic. This acknowledges that different use cases may require different approaches to model availability and governance.

Teo offered a compelling anecdote of a small Singaporean eatery leveraging an AI assistant to analyze past sales records and optimize future promotions. "I can ask it based on my past sales record to tell me which of my past promotions have been more successful," she quoted the owner. This demonstrates that practical AI applications, even those not requiring direct GPU access, can deliver tangible business benefits, dispelling the notion that AI is solely for tech giants.

Related Reading

• Executives See Delayed But Inevitable AI Job Impact
• AI Chip Spending: The Looming Depreciation Cliff
• AI's Super Cycle: Incumbents Lead, Disruption Looms

The dialogue also touched on education, highlighting its transformative potential. Ibrahim shared a pilot program in Northern Ireland where 100 teachers, utilizing AI tools over six months, reported an average saving of 10 hours per week. This productivity gain allows educators to focus on their core passion: transferring knowledge and engaging students. Teo advocated for personalizing learning, leveraging AI to cater to individual strengths and weaknesses, fostering better learning outcomes. This isn't just about layering AI onto existing systems; it's about fundamentally rethinking how education can be delivered more effectively and equitably.

The discussion concluded with a recognition that the digital divide in AI is a multifaceted problem, demanding a blend of technological innovation, strategic policy, and international cooperation. It is a continuous process of adaptation and learning, where fostering human capabilities and ensuring equitable access to foundational tools remain paramount.