NEXT Frontier, What is Sovereign AI

Sovereign AI is a key enabler for a nation's capability in the world of artificial intelligence (AI). It revolves around a country's ability to independently create and utilise AI technologies using its own resources.

Consider a country with its own clandestine AI lab, where brilliant minds collaborate to develop groundbreaking AI solutions. This lab is not just about sophisticated gadgets and computers; it thrives on the country's infrastructure, data, skilled workforce, and business connections.

In simpler terms, Sovereign AI entails a country's capability to rely on itself for AI development and usage, without external dependencies. The concept is likened to having an expert team of national crusaders address the key challenges of the digital era.

This AI concept transcends technology; it is about a country's identity and autonomy. It ensures that AI advancements align with the nation's values and culture, maintaining authenticity in the rapidly evolving landscape of AI innovation.

What are some of the AI Model Categories?

Explore AI model categories: Public, Sovereign, and Private. Learn their features and functions to understand how they shape the AI landscape.

1. Public AIs

• These AI models are developed and determined by external entities.
• They serve as universal tools accessible to all users.
• Public AIs undergo training using vast amounts of publicly available or proprietary data. Service providers have access to and store this data, including both training and inference data, to continually enhance the AI model's performance through user input.
• Data has the potential to move across the internet. 
• Examples: Google's BERT (Bidirectional Encoder Representations from Transformers), OpenAI's GPT (Generative Pre-trained Transformer).

2. Sovereign AI (Open-source offshoots & multi-tenant AIs)

• This category comprises self-determined AI models.
• Sovereign AI models in this category can integrate AI capabilities while maintaining control over the data they utilise.
• These Sovereign AI models offer the advantage of adaptability across diverse regions, including Asia Pacific, Europe, and the Americas. They can operate proficiently in multi-tenant environments, accommodating regulatory requirements such as the EU Data Boundary regulations, making them suitable for deployment in various jurisdictions worldwide.
• This flexibility allows corporations to leverage external resources and expertise while still upholding their data governance preferences and jurisdictional requirements.
• These models can be customised and fine-tuned for specific applications or tasks.

• Example: Hugging Face's Transformers library, Microsoft's OpenAI GPT models available in Azure, LLama2.

3. Private AI (Self-sufficient)

• Self-sufficient Private AI models have their own proprietary AI algorithms developed by a third party or internally and the architectures is hosted in a secure private on premises or colocation environment.
• They prioritise maximum control over their operations, as these algorithms tend to contain highly sensitive and valuable data.
• These models operate independently on local systems without relying on external parties, and data travels on private, secure network connections. Private AI focuses on retaining ownership and control over your company's data and intellectual property. Instead of outsourcing it to external commercial entities, where security measures and data handling practices may be uncertain.
• The Private AI modesl re highly specialised for specific tasks and functions, tailored to meet the unique needs of the organisation or user.
• Example: DeepMind's AlphaGo, Facebook's RoBERTa (Robustly optimized BERT approach)

Why is Sovereign AI Important?

Sovereign AI empowers nations to cultivate their AI capabilities using their own resources, promoting innovation and fostering sustainable growth. By leveraging indigenous infrastructure, data assets, skilled talent pools, and collaborative business ecosystems, countries can chart their path towards technological self-sufficiency and global competitiveness.

In essence, Sovereign AI represents a nation's journey towards autonomous innovation, shaping the landscape of AI development and deployment on a global scale.

Industry leaders, including Nvidia CEO Jensen Huang and IBM Corp. CEO Arvind Krishna, have echoed the importance of Sovereign AI. Huang stressed the criticality of countries retaining control over their AI infrastructure to preserve their cultural identity. He emphasized that Sovereign AI isn't solely about technological advancement but also about safeguarding a nation's unique cultural, societal, and historical characteristics.

Krishna further emphasized the significance of every country having sovereign AI capabilities, advocating for the establishment of national AI computing centers and standardized data sets. These statements underscore the global momentum towards embracing Sovereign AI as a fundamental pillar of technological progress and national sovereignty.

The rapid evolution of AI is opening up vast opportunities for economic growth, problem-solving, scientific breakthroughs, and bolstering national security. Predictions indicate that AI could drive a whopping $20 trillion in global economic value by 2030.¹ This immense potential has sparked a global race towards AI development. As of December 2020, over 30 countries and regions had already crafted strategic plans for AI, while top-tier universities ramped up investments in AI education.² Simultaneously, private sector funding in AI ventures has surged, reflecting the widespread acknowledgment of AI's transformative power and its pivotal role in shaping our collective future. Notably, in Australia, the Department of the Prime Minister and Cabinet has recognized AI and machine learning as vital technologies essential to national interests.³

References

(1) SA Hajkowicz, S Karimi, T Wark, C Chen, M Evans, N Rens, D Dawson, A Charlton, T Brennan, C Moffatt, S Srikumar, KJ Tong, Artificial intelligence: solving problems, growing the economy and improving our quality of life, CSIRO Data61, Commonwealth Scientific and Industrial Research Organisation, Australia, 2019, online.

(2) Daniel Zhang, Saurabh Mishra, Erik Brynjolfsson, John Etchemendy, Deep Ganguli, Barbara Grosz, Terah Lyons, James Manyika, Juan Carlos Niebles, Michael Sellitto, Yoav Shoham, Jack Clark, Raymond Perrault, The AI Index 2021 annual report, AI Index Steering Committee, Human-Centered AI Institute, Stanford University, Stanford, California, March 2021, online.

(3) Critical Technologies Policy Coordination Office, The Action Plan for Critical Technologies, Department of the Prime Minister and Cabinet, Australian Government, 2021, 15–16, online.