It appears that accessing large quantities or advanced compute resources are no longer the defining and sustainable advantage that many had expected. The capability gap between the leading US and Chinese models is essentially gone, and the Chinese models now have a distinct advantage in that they are able achieve Nearly equivalent resultswere achieved using a fraction of the computing resources available to leading Western labs.
The AI contest is increasingly framed in national security terms as a zero sum game and influenced by the assumption that a future conflict between the US and China centered around Taiwan is inevitable. The US has used ahref=””https://www.cnn.com/2023/10/18/tech/us-china-chip-export-curbs-intl-hnk/index.html””>”chokepoint]tacticsin order to limit China’s ability to access key technologies such as advanced semiconductors. China has responded to this by accelerating their efforts towards self-sufficiency, which has caused US efforts to backfire.
Recently even outgoing US Secretary of Commerce Gina Raimondo, a staunch advocate for strict export controls, finally admitted that using such controls to hold back China’s progress on AI and advanced semiconductors is a “fool’s errand.”Ironically, the unprecedented export control packages targeting China’s semiconductor and AI sectors have unfolded alongside tentative bilateral and multilateral engagementsto establish AI safety standards and governance frameworks–highlighting a paradoxical desire of both sides to compete and cooperate.
If we examine this dynamic in more detail, it becomes clear that there is no real existential danger from China. Instead, the real threat comes from rogue groups and bad actors who want to weaponize advanced AI for their own gain, to destabilize society, or to cause widespread harm. As with nuclear weapons, China as a nation state must be cautious about using AI-powered capability against US interests. However, bad actors, such as extremist groups, are more likely to abuse AI abilities without hesitation. It is difficult to defend and prevent a determined foe from using AI technology for malicious purposes, given its asymmetrical nature, similar to cyberweapons.
In light of the implications, it is incumbent upon the US and China, as global leaders in AI technology, to jointly identify and mitigate these threats, collaborate on solutions, cooperate on developing a framework for regulating most advanced models, and work together on developing a global regulatory framework. This is instead of erecting fences around AI technologies, no matter how large or small, and pursuing policies that divert attention from the real danger.
Despite the high stakes, escalating rhetoric and the current course of intense competition, it is clearer than ever before that there will be no long-term winners. The consequences could be grave, undermining global stability and stalling scientific advancement, as well as leading both nations to a dangerous technological brinkmanship. This is especially relevant given the importance Taiwan and TSMC, the global leader in foundries, has on the AI stack and the increasing tensions surrounding the high-tech Island.
If we continue on this path blindly, we risk isolation and polarization. This will threaten not only the international peace, but also the vast benefits AI can bring to humanity.
The current state of US-China AI rivalry is a result of historical narratives, geopolitical factors, and economic competition. Recent reportby the US-China Economic and Security Review Commission frames the issue in binary terms. It is centered on dominance or submission. This “winner-takes-all” logic ignores the potential of global collaboration and could even cause a self fulfilling prophecy through escalating conflict. This dynamic will likely be accentuated under the new Trump administration, with increased discussion of a Manhattan Projectfor AI and redirection US military resources away from Ukraine towards China.
Fortunately a glimmer hope for a responsible AI collaboration is emerging now, as Donald Trump posted on January 17, that he had restarted direct dialogregarding various areas of cooperation, and given previous cooperation should continue to “partners and friend.” The TikTok drama will be a preview to how his efforts to set US China relations on a more confrontational trajectory.
AI’s promise for good
Western media usually focuses attention-grabbing topics described as “existential risks of bad AI.” Unfortunately, AI safety experts who receive the most coverage often recite similar narratives, scaring public. In reality, there is no credible evidence that a more powerful AI will become more evil. We must challenge the false dichotomy between pure accelerationism and doomerism in order to create a model that is more like collaborative acceleration.
There is a significant difference in the way AI is perceived by developed and developing countries.In developed countries, the public’s perception of AI is negative (60-70%) while in developing markets, the ratings are positive (60-80%). The people in the developing markets have seen how technology has transformed their lives in the last few decades. They are hopeful that AI will help them solve the remaining problems they face by improving their education, health care and productivity. Western populations are often unaware that these same benefits can directly improve their lives, given the high levels inequity in developed markets. Imagine the progress that could be made if we redirected the trillions of dollars that are spent on defense budgets every year to infrastructure, health-care, and education projects.
When we reach the next phase of AI, it will help us accelerate our scientific discoveries, develop new drugs and extend our lifespan. It will also reduce our work commitments, and ensure that everyone has access to a high-quality education. It may sound like a pipe dream, but current trends suggest that most of these things can be achieved within a decade, and even sooner. To get there, we’ll need to develop more advanced AI systems. This will be a more challenging goal if compute/data resources are divided and research talent pools are not. According to industry studies, almost half of the world’s top AI researchers (47%) areChinese born or educated in.Without the efforts of Chinese researchers, it’s difficult to imagine where we would be today. Active collaboration with China in joint AI research can be pivotal for supercharging progress.
As the AI competition between China and the US intensifies, it poses serious threats to both countries and the entire world. These risks are not hypothetical. They could have consequences that threaten global stability, economic peace, and technological advancement. Framing artificial intelligence development as a zero sum race undermines the opportunities for collective advancement and safety. It is vital that the US, China, and their allies shift away from the rhetoric of confrontation and toward collaboration and shared governance.
We have some recommendations for policymakers.
- Lessen the influence of national security on AI policy.The US and China need to re-calibrate their approach towards AI development. They must stop viewing AI as a purely military tool. This means that national security concerns, which currently dominate AI policy, should be less of a focus. Instead, policymakers need to focus on civilian AI applications that can directly benefit the population and address global issues, such as climate change, health care, and education. The US should also investigate how to implement an possible universal basic income program, as job displacement due to AI adoption becomes more of a domestic issue.
- 2.Encourage bilateral and multilateral AI Governance.A robust dialogue between the US and China, as well as other international stakeholders, is essential for the development of AI governance standards. This includes establishing ethical norms, safety guidelines, and transparency guidelines regarding advanced AI technologies. A cooperative framework could help ensure that AI is developed responsibly and inclusively while minimizing risks and maximising benefits for all.
- 3.Increase investment in detection and prevention of AI misuse.The biggest existential risk to the world is the risk of AI misuse, whether it be through misinformation campaigns or cyber attacks that could destabilize the society. It is crucial to dramatically increase funding and international cooperation for detecting and mitigating risks. The US and China need to agree on standards for responsible AI use and work together on tools that can monitor misuse and counter it globally.
- 4. Create incentives to encourage collaborative AI research.The government should provide incentives to encourage academic and industrial collaborations across borders. By creating joint research and funding initiatives, the US can create an environment in which the best minds of both nations can contribute to AI breakthroughs that benefit humanity as a group. This collaboration will help pool talent, data and computing resources, overcoming obstacles that neither country can overcome alone. A global effort similar to the CERN for AI (19459020) will bring more value to the world and a peaceful ending than the Manhattan Project for AI (19459020), which is being promoted today by many in Washington.
- 5.Establish measures to build trust.Both nations must prevent AI-related actions from being misinterpreted as aggressive or threatening. This could be done through data-sharing agreements and joint projects in nonmilitary AI. They could also exchange AI researchers. Reduced import restrictions on civilian AI applications, for example, can help nations rebuild trust and allow them to discuss a deeper level of cooperation. These measures would increase transparency, reduce the chance of miscommunication and pave the path for a more adversarial relationship.
- 6. Support the creation of a global AI Safety Coalition.The formation of a coalition with major AI developers from different countries could be a neutral platform to address ethical and safety issues. This coalition would bring together leading AI scientists, ethicists, policymakers, and other experts to ensure that AI advances in a fair, safe, and beneficial way for all. This effort should include China, which is an important partner in the development and maintenance of a safe AI eco-system.
- 7.Focus on AI to solve global challenges.The two AI superpowers must use their capabilities to address global issues such as climate changes, disease and poverty. By demonstrating AI’s positive social impacts through tangible projects, and by presenting AI not as a menace but as a powerful force for good, both the US and China could reshape the public perception of AI.
The choice is clear and simple: we can continue down the path of confrontation, which will almost certainly result in mutual harm, or pivot towards collaboration, which has the potential to create a prosperous, stable future for everyone. Artificial intelligence has the potential to solve many of the world’s greatest challenges. But realizing that promise depends on our decision to work together or race against one another.
The world cannot afford to pass up the opportunity to harness AI to the benefit of all.
Alvin Wang Graylin.
Alvin Wang Graylin, a technology executive and investor with over 30 years’ experience in shaping innovation, is a pioneer and innovator who has shaped innovation in AI, XR, cybersecurity and semiconductors. Graylin is currently the global vice president of HTC. He was previously the company’s China President from 2016 to 2023. He is the authorOur Next Reality ().
Paul Triolo (
Paul Triolo]is atechnology policy lead anda partner for China at DGA-Albright Stonebridge Group. He advises clients from the technology, financial service, and other sectors on complex political and regulatory issues in the US, China and the European Union.
