From generative language models to GPTs, today's most prominent AI tools are owned by private companies. Such concentration poses the risk that these platforms will shape global culture by amplifying biases or perpetuating injustice.
The State of Schleswig-Holstein and Nextcloud have joined forces to develop digitally sovereign AI technology that meets European data privacy, security and digital sovereignty regulations - an integral component of its AI strategy.
Will Your Nation Be Left Behind?
Only a select group of AI companies control most of the world's powerful AI technology, providing an incredible advantage in terms of innovation and competitiveness, but placing all other people at risk. Since AI involves not just big data and algorithms but also requires hardware and software solutions that deliver on its promises, Sovereign AI provides governments with tools they need for specific applications.
Sovereign AI tools offer an alternative to the current ecosystem dominated by big tech-centric AI. Generative AI requires access to personal data for training purposes, which could potentially be exploited for commercial gain or malicious intent by third parties. Sovereign AI solutions offer better privacy and security controls to minimize these risks for safer, trustworthy, and more effective generative AI apps.
Sovereign AI has gained considerable momentum globally. India, for instance, has highlighted the need to develop its own AI capabilities as part of economic development initiatives. They plan on doing this by taking advantage of Aadhaar as a governance technology and creating localised digitally sovereign AI platforms.
Australia, too, is taking steps to develop its AI infrastructure by investing in education and hiring skilled staff. As demand for advanced AI applications rises, nations' ability to leverage their own AI infrastructure will become ever more crucial.
Indeed, many international companies are already creating their own private and scalable AI platforms. Del Complex provides one such solution, providing a tight-guarded barge dedicated to AI research that operates internationally without needing permits for operation.
Though the risk of human beings losing control of artificial super intelligences remains very real, a balanced perspective must still be kept in mind. The problem goes far beyond one unethical AI developer, social media platform or cybersecurity agency; rather it involves taking a whole-of-society approach combining law-based governance (risk evaluation, oversight of accountable exemptions standards auditing etc) with technical requirements of developing AI software/hardware solutions.
5 Reasons You Need a Sovereign AI Strategy
As AI competition heats up, nations are pushing harder for the right to develop their own sovereign AI capabilities. This trend has gained steam thanks to Generative AI (GAP) models such as ChatGPT and DALL-E from OpenAI; Bard from Google; Claude from Anthropic; and Llama 2 from Meta - which enable nations to develop these capabilities independently.
GAPs represent an emerging form of artificial intelligence that differs significantly from the standard machine learning (ML) and deep learning models currently dominating AI buzz, in that they can be trained on diverse sets of data in order to generate results specific to an environment, language and cultural context. They could potentially solve many difficult issues that were once impossible or impractical using this approach.
But developing AI applications that run on your own national infrastructure requires significant investments of talent and computing power, according to Macon-Cooney. Governments may find themselves unable to compete with private sector companies that provide higher salaries and more opportunities for innovation.
These applications must access personal or proprietary data to train on, which raises further privacy and security issues. A sovereign AI model would give greater control over these assets to prevent unauthorized access as well as cross-border transfers, monetization or any violations to sovereignty.
And finally, sovereign AI provides the chance to incorporate your values and beliefs into the model itself, according to Luccioni. She notes that each model encodes values; so it is crucial that any models we build incorporate our individual norms and beliefs as part of their foundation.
Even with its challenges, a sovereign AI approach makes sense in certain sectors of the economy. But this should not be seen as global harmonization: individual countries differ in their priorities, legal traditions, economic structures, demographics and geography; therefore robust international cooperation that centers around shared values may help even out the playing field and lead to the development of AI tools that meet everyone's needs.
7 Steps to Building an AI-Powered Future
Building an AI-compliant architecture that unlocks its benefits while meeting national laws' compliance requires an intricate partnership between law-makers and technologists.
Luccioni notes that any partnership must be both global and local in nature. It should focus on identifying key challenges related to data residency and cross-border transfers that pose data sovereignty risks to an AI application; while also considering how its solution will fit within local legal systems in terms of meeting data classification laws such as those found in Schrems II for EU citizens.
So it is vital that countries create their own autonomous AI strategy. Instead of relying on large language models developed by OpenAI and Anthropic or their computing infrastructure to support them, nations should invest in their own capabilities - something which would only cost millions rather than billions - so as to leverage AI's power for economic development purposes.
An AI strategy designed for sovereign countries should also help ensure that AI applications understand the context in which they operate, including cultural values and social norms. To do this, AI solutions developed and deployed should take into account national identity, political systems and regulatory structures as well as business processes - this way ensuring users get useful and valuable information while abiding by applicable laws and regulations in each jurisdiction in which the model operates.
Finally, any sovereign AI initiative must be driven by a clear public purpose to ensure it will be accepted by its target community and successfully implemented within an organization. Furthermore, this will help ensure the technology is used ethically as well as legally.
The Ethical AI Imperative
Children learn that acting simply because everyone else does it may not be wise. We should take our own advice as adults, avoiding temptations to follow popular AI trends that could lead us down paths of social or moral destruction. It's crucial that businesses using AI as part of their innovation strategies take time to consider its wider repercussions before acting unilaterally on them.
Ethical AI requires taking an inclusive, explainable, and beneficial approach to designing AI systems. This includes having full knowledge of all data used to train the model as well as conducting an audit to ensure there is no inherent bias or other problematic attributes present within its algorithm. Furthermore, ethical AI ensures that third parties do not attempt to manipulate it for their own ends.
Education must be the cornerstone of this effort by all stakeholders - from executives and data scientists, frontline employees and consumers, educators as well as front-line managers and consumers. Education on policies, key considerations, potential negative repercussions of data sharing and AI automation is essential - university students alone cannot create education materials solely geared toward AI ethics issues - rather it must address ethical considerations across the board.
To address these concerns, several initiatives have been introduced to foster an ethical AI environment in business. One such effort is Aleia - an umbrella organization which unifies European AI startups into one single resource for corporate clients. Backed by an arm of France's government as well as angel investors, its goal is to strengthen European technology sovereignty by providing more secure and reliable business-critical AI solutions.
Schleswig-Holstein in Germany is taking an even more radical approach. Partnering with Nextcloud, an open cloud collaboration platform, it recently unveiled an initiative called "Digital Sovereignty Through Artificial Intelligence." This will enable its government to use an AI assistant embedded within Nextcloud that only has access to its encrypted data and cannot access other state datasets.
No comments:
Post a Comment