How to free the future of AI? Taiwan Model Can Help
The world is perched on an artificial intelligence precipice. But how to avoid tumbling into an abyss of abuse, exploitation and manipulation by bad actors in the pay of those seeking to extinguish freedoms, subvert democracy, threaten human rights and undermine rules-based order?
Audrey Tang is Taiwan's minister of digital affairs and chair of the National Institute of Cyber Security.
This critical question dominated discussions with numerous innovative public and private sector minds during my recent visits to Belgium, Israel and the U.K. Also on the tip of everyone’s tongue was the pressing need to cooperate on AI.
There is a smorgasbord of strategies for advancing this exalted state of being. These include a pause in training next-generation AI systems, industry-crafted rules and standards, government regulation and everything in between. But to my mind, the Taiwan Model is best suited for AI global governance and worth sharing far and wide.
We are no strangers in Taiwan to thinking outside the box when it comes to building consensus on all issues great and small. It is imperative, given our geopolitical situation, for the people to feel fully invested in policymaking processes. This strengthens cohesion and reflects our deep-seated belief that more democracy, not less, is the best way forward.
The Taiwan Model, which recognizes the power of AI as a force for good, uncovers opinions and perspectives on an array of issues. The end goal is promoting transparency and moving beyond division to create consensus. This reflects our passion for freedom, inclusivity and fairness forged on the front lines of authoritarian expansionism.
One of the pillars of the Taiwan Model comprises open source platforms like the homegrown Trustworthy AI Dialog Engine (TAIDE). The chatbot, which recognizes the importance of linguistic diversity and quality, answers general knowledge questions, fast-tracks translation processes and supports industries with reliable and up to date information.
The other pillar, Alignment Assemblies, sets the Taiwan Model apart. Piloted by Collective Intelligence Project (CIP), alongside partners such as Taiwan’s Ministry of Digital Affairs, OpenAI and Anthropic, it aims to bring technology in sync with societal values. We reject arbitrary top-down directives in favor of a more inclusive, participatory model, shaped by our vibrant civil societies.
Commencing this month via a series of internationally staged Alignment Assemblies, we are addressing a host of under-the-radar issues like copyright, bias and discrimination, due compensation, fair use, public service and broader societal impacts. The process is powered by Polis - a real-time system capable of gathering, analyzing and understanding what large groups of participants think. Its allure lies in a simple yet profound design: People naturally gravitate toward finding common ground, rather than delving into divisive issues.
The truth of the matter is open source platforms and Alignment Assemblies are potent catalysts for deepening understanding and unity. When the people see and touch the fruits of collaboration, the result is co-creation and solidarity. This virtuous cycle is key to revitalizing democracy, ensuring we leave no one behind in tackling existing and emerging challenges to our free and open societies.
High on this list is climate justice. Although not specifically mentioned in the Statement on AI Risk by the Center for AI Safety, which I signed along with a legion of leading experts and public figures, it deserves to be.
After a recent dialogue of depth with Simon Kofe, Tuvalu’s minister of justice, communication and foreign affairs, as well as a global champion of raising climate change awareness, I am convinced we must confront the devastating impact of surging sea levels on Pacific nations like the Marshall Islands, Nauru, Palau and Tuvalu, to name but a few. The Taiwan Model, in this case, can help.
Consider the issue within the context of potential short-term harms of AI like interactive deepfakes. If this scourge is allowed to unpick the tapestry of trust binding us together, cross-border coordination on global threats such as the climate crisis is impossible, pushing us closer to catastrophe.
The bottom line is it takes all hands on deck to cut the AI Gordian knot whilst charting a safe and sustainable course for sound global AI governance. This is why I invite Mykhailo Fedorov, Ukraine’s minister of digital transformation, to join the CIP partnership and continue our April 2022 collaboration when we were among representatives of over 60 countries to sign the Declaration for the Future of the Internet.
In fact, all public, private and nongovernmental organizations in Ukraine sharing our vision for the future of AI are welcome to participate. By amplifying a plurality of voices, as well as harnessing the synergy of the Taiwan Model for the collective good, we can minimize risks, maximize benefits and free the future—together.