As U.S. President Donald Trump prepares to visit Beijing on May 14-15, 2026, for a highly anticipated summit with President Xi Jinping, the world is watching to see if the two leaders can stabilize a U.S.-China relationship strained by disputes over trade, technological race, the future of Taiwan, and the rippling effects of the conflict with Iran.

Trump’s trip to Beijing – already rescheduled once due to the conflict in the Middle East – has been described as having tremendous symbolic significance. Yet, expectations for a breakthrough on specific deliverables should remain low, according to Susan Thornton, a China expert and former U.S. diplomat. Thornton joined APARC Director Kiyoteru Tsutsui on the latest episode of the APARC Briefing video series to analyze the potential outcomes of the Trump-Xi summit and the high-stakes dynamics shaping U.S.-China relations.
 

Image

Trade and Tech: A Call for a Paradigm Shift

On the economic front, Thornton drew on her deep experience in trade negotiations to critique the lack of focus in U.S. policy.

"The biggest problem for U.S. negotiators is deciding what it is that we want from China," she said. "We tend to give them a long list of revolving priorities, which [makes it easy for the] other side of the negotiating table to just fob them off and not actually commit to anything over years of negotiations.”

On the technology rivalry between the two powers, Thornton urged a shift in strategy. Rather than pursuing sweeping export controls that are often unilateral and incomplete, she advocated for a narrower, multilateral approach focused on the most sensitive technologies, combined with a greater emphasis on American innovation. AI governance is one of the areas Thornton believes could be a common ground for Washington and Beijing to align their policies.

“It's going to be very hard for the United States to contain China's technological ambitions and growth,” she said. “I don't think that we're exactly competing on the same metrics. I question how it is that we're going to be able to keep China from getting technologies that are dual-use but might be useful in some military application when these things are basically economy-wide products.”

When it comes to technological competition, "We need to try to run faster than China, not be constantly trying to trip China up and looking in the rearview mirror," Thornton urged. "I don't think that's going to bode well for the long-term development of the U.S. tech sector."

The Taiwan Flashpoint: A Longer-Term Challenge

While Taiwan remains the most dangerous flashpoint that could trigger a kinetic warfare between the United States and China, Thornton believes that the immediate risk of conflict has receded, in accordance with recent U.S. threat assessments that no longer see 2027 as a likely target date for a potential Chinese takeover of the island.

Beijing, she argued, is closely watching the domestic political situation in Taiwan and how the leadership in Taipei views U.S. reliability and support. “I think the Chinese have determined, based on both of those things they've been watching, that they can afford to wait a bit longer, see what happens.”

Thornton cautioned, however, that, even as a conflict over Taiwan may no longer pose an immediate-term threat, “it is a problem that is going to develop over the coming decade.”

Diplomacy in a Multipolar World Order 

When asked about the future of the global order, Thornton described a trend toward fragmentation. If the United States steps back from its global leadership role, it is difficult to see who else would be willing or able to shoulder the cost of providing global public goods, she said. A “thinner world order,” with the United Nations at its center, may eventually find favor with countries that can afford to pay for some of those goods, she reflected.

In a closing advice for aspiring foreign service officers, Thornton argued that the emergence of a multipolar world reinforces the need for skilled diplomacy. “As the global order changes and more countries come into the mix of the councils of politics in the world, the United States will have to lean back toward diplomacy more,” she predicted.

“We're going to need very good diplomats,” she concluded, because it will be significantly harder to be an American diplomat in a fragmented world order in which the United States is no longer the single overwhelmingly dominant power.

Introduction and Contribution:

Wars have wide-ranging effects on the political attitudes and behaviors of citizens and elites. For example, European leaders were pressured to make democratic reforms and build large welfare states during World War II in order to stabilize their countries and encourage wartime sacrifice. After the 9/11 attacks and the initial Afghanistan invasion, George W. Bush’s approval ratings soared to 90%.

The Russo-Ukrainian War has been one of the defining wars of this century. Its consequences for the politics of both countries — such as Ukrainian national unity or the repression of Russian dissidents — are beginning to be understood. Yet it is less clear how the war has affected third-party states. In light of Russia’s imperial ambitions, as well as Ukraine’s need for international support to combat a regional hegemon, this is a pressing issue.

Leaders of third-party states have responded to the war in a range of ways, including imposing sanctions on Russia and mediating between the two countries. Wars also affect how rulers communicate with their citizens: they may be motivated to emphasize their shared connections with either warring party, their need for national self-reliance in a hostile international system, and so on.

In “Between wars and words,” Ana Paula Pellegrino, Benjamin R. Burnley, and Laia Balcells show that leaders of third-party states increased their nationalist rhetoric on X (formerly known as Twitter) after the invasion. The authors analyze over 10,000 tweets from the heads of 130 states both before and after the invasion, mapping these along a nationalist-cosmopolitan spectrum. The effects of Russia’s invasion on nationalist tweets were strongest among North Atlantic Treaty Organization (NATO) members and weaker among members of the pro-Russia Collective Security Treaty Organization (CSTO).

“Between wars and words” provides evidence that leaders react to wars in similar ways as do masses, whose sense of national identity tends to increase during periods of global uncertainty and conflict. As politicians increasingly use X to communicate, a more precise understanding of their tweets before and after major conflicts may help social scientists better understand their beliefs and threat perceptions. 

Methodology:

The authors collect tweets from the 14 days before and after the invasion using an unsupervised learning algorithm called GloVe. This process is diagrammed in Fig. 1 below. GloVe identifies relationships between words by analyzing how often they appear together, grouping together words with similar meanings. This enables the authors to code each tweet according to its topic and sentiment (e.g., positive vs. negative).
 

Image

To code the level of nationalism versus cosmopolitanism, the authors employ the Concept Mover’s Distance (CMD) approach. CMD measures how closely the words in each tweet align with sets of ‘anchor terms’ representing each concept, placing tweets along a continuum between the two. Briefly, nationalism refers to a set of beliefs about the reality and value of nations, the obligations that members of a nation have to one another, and the right of that nation to determine its political affairs, whether as its own state or within an existing state. By contrast, cosmopolitanism is the view that all humans, irrespective of their national memberships, ought to be seen as part of a single world (kosmos) community. Returning to CMD, nationalist tweets are those that align with anchor terms such as ‘pride, glory, patriot, forefathers, homeland,’ and ‘heritage.’ Cosmopolitan anchors include ‘cooperation, humanity, multilateral,’ and ‘universal.’ 

Importantly, nationalistic tweets need not bear an obvious relationship to (inter)national security — the authors hypothesize that Russia’s invasion merely increased nationalist rhetoric, not rhetoric of a specific kind. For example, 10 days after the invasion, Bolivian President Luis Alberto Arce Catacora tweeted, “In 2019, the glorious Alteño people once again showed us their courage and love for our country.” Conversely, 12 days prior to the invasion, Argentine President Alberto Fernández espoused several cosmopolitan sentiments when he tweeted: “With the logic of multilateralism, Argentina has discussed with Russia the possibility of deepening financial assistance and increasing bilateral investment and trade between the two countries.” 

Findings:

The authors show that Russia’s invasion did significantly increase nationalist rhetoric by third-party heads of state, whether or not one controls for topic and sentiment. As an example of this, consider Austrian Chancellor Karl Nehammer: Three days prior to the invasion, he tweeted, “Austria continues to rely on diplomacy and de-escalation to prevent war. The OSCE…is the appropriate framework.” On the day of the invasion, by contrast, Nehammer tweeted “I promise that I will do everything to protect the people who live in Austria.” Other international leaders, such as Joe Biden and Justin Trudeau, tweeted out more cosmopolitan statements about the importance of collectively pressuring Putin and respecting international law.
 

Image

Why might an international war prompt the leaders of third-party states to use more nationalist rhetoric? The authors provide a number of hypotheses. For one, wars lead to feelings of uncertainty, and nationalist rhetoric by elites can afford citizens a sense of safety. Third parties might also make nationalist statements to signal their support of (or opposition to) one of the belligerent parties; for example, “our proud nation will not stand idly by as Russia attacks Ukraine.”

Although the invasion increased nationalist discourse on average, these effects are driven primarily by the behavior of NATO members. This is likely because the alliance has historically taken a strong stand against Russian aggression. As mentioned above, CSTO members did not tweet in a more nationalistic way, nor did leaders of states with histories of territorial armed conflicts resembling the Russo-Ukrainian War.
 

Image

*Brief prepared by Adam Fefer.

As people increasingly turn to large language models for political tasks, including voting guidance, the political neutrality of AI chatbots has emerged as a major policy concern. American AI chatbots are used globally, yet little is known about their behavior as tools for political decision-making and potential political bias in non-U.S. contexts.

To address this gap, researchers ran an experiment during the final week of Japan’s February 8, 2026, general election. The experiment reveals a striking pattern: when asked which party to support in the election, five major AI models from three companies overwhelmingly directed voter profiles with left-leaning policy positions toward the Japanese Communist Party (JCP). The reason, according to the researchers, has to do with the information environment AI systems can access.

These findings, published in a working paper titled Why Do AI Models Tell Left-Wing Voters to Support the Communist Party?, “suggest that AI voting advice may be shaped as much by the information-retrieval environment as by model training, with implications for governance frameworks that rely on U.S.-centric assumptions,” write the researchers, Andrew Hall, the Davies Family Professor of Political Economy at the Stanford Graduate School of Business and a senior fellow at the Hoover Institution, and Sho Miyazaki, a visiting researcher at Waseda Institute of Political Economy, an incoming Ph.D. student in public policy at Harvard University, and a former predoctoral research fellow at Stanford University. Miyazaki is also a core member of the Stanford Japan Barometer, a project of the Japan Program at Stanford’s Shorenstein Asia-Pacific Research Center.

Sign up for APARC newsletters to receive our scholars' research updates >

How AI Models Deliver Political Advice in Japan: A Systematic Experiment

To understand how AI models provide political recommendations in the Japanese context, Hall and Miyazaki created 36,300 synthetic voter profiles with varying gender, region, and stated political views on 12 policy issues spanning security (constitutional amendment, defense spending, espionage law), diplomacy and immigration (China relations, foreign workers, permanent residency), energy (nuclear power), economic (consumption tax, social insurance), and social domains (dual surnames, restrictions on corporate donations, Diet seat reduction).

They then queried five models from three AI companies (OpenAI, Google, and xAI) during Japan’s February 8, 2026, Lower House election, asking each model to recommend a political party based on the voter profiles. All five models were queried with web search enabled and could access current information.

The researchers found that policy positions overwhelmingly dominated the models' party recommendations, producing swings of 50 to 98 percentage points in party choice, compared to just 0.5 to 7 percentage points for demographic factors. Thus, demographic effects are an order of magnitude smaller than policy effects.

Furthermore, left-leaning policy views in voter profiles caused all five AI models to converge overwhelmingly on recommending the Japan Communist Party, even though other parties hold broadly similar positions on the issues tested. The concentration on recommending JCP under left-leaning policy stances is therefore not explained by ideological distinctiveness.

In the control condition without policy input, models showed no uniform left-wing bias: three of the five models recommended the Liberal Democratic Party at high rates, and JCP shares were low for four of the five models.

“The key finding is that JCP recommendation rates rise sharply when policy positions are provided, which is the typical scenario when voters use these tools in practice,” write Hall and Miyazaki.

Information Environment Asymmetry

Why the JCP? The researchers traced the pattern to the sources AI models cite when making recommendations.

The JCP operates Akahata, a self-described daily newspaper published on a fully open website that AI web-search tools can freely access. In contrast, Japan's major news outlets have implemented technical barriers (known as robots.txt restrictions) that block AI crawlers from accessing their content, a move driven by copyright concerns.

The researchers found that the JCP's open website and party newspaper were among the most-cited sources in the AI models’ recommendations. Unable to distinguish between editorially independent journalism and partisan content, the models treated the JCP content as a credible news source. Thus, the information environment available to AI is systematically skewed toward the JCP's partisan sources that are designed to persuade rather than to scrutinize and inform.

“A model that retrieves information from jcp.or.jp/akahata and simultaneously classifies that site as news media is not simply making a labeling error: it is operating in an information environment where the boundary between party communication and journalism is genuinely blurred, and where the consequences of that blurring flow directly into its recommendations,” Hall and Miyazaki write.

The researchers also found that incorporating X search amplified left-leaning recommendations in Japan, the opposite of expectations based on the U.S. discourse environment.

Implications for Democratic Systems in the AI Age

The study's findings carry significant implications:

• AI governance frameworks should treat content access policy and AI political neutrality as deeply intertwined domains.
• Election commissions should create nonpartisan platforms that compile structured data about party positions so that the information is comparable, party-independent, and machine-readable.
• News organizations should recognize that by imposing copyright-motivated content access restrictions, they may inadvertently cede influence over AI-mediated information to partisan actors. They may wish to consider forms of negotiated access.
• Political actors will likely begin to optimize their communication for AI.
• Users should exercise caution in using AI as a voting advisor and be conscious of its potential biases and blind spots.

“If AI systems are going to act as political intermediaries more broadly, two problems need to be addressed,” writes Hall in an article about the research via his Substack. “The first is informational: ensuring that what the sources models read reflects the same balance of scrutiny and debate that voters encounter in a healthy media ecosystem. The second is advisory: deciding how an AI system should even translate a voter’s values into political guidance in the first place.”

Learn more about the Stanford Japan Barometer and its work  >

How did Taiwan transform itself from an impoverished economy into a titan of computer and semiconductor chip manufacturing, home to industry giants such as Acer, Foxconn, and TSMC?

The history presented in Honghong Tinn's recent book, Island Tinkerers: Innovation and Transformation in the Making of Taiwan’s Computing Industry (MIT Press), demonstrates that Taiwan's successful computing enterprises and manufacturing sectors arose from Taiwanese technologists' engagement in a process of technology transfer involving overlapping acts of imitation, emulation, experimentation, and innovation.

Tinn, a historian of information technology based at the University of Minnesota Twin Cities, discussed her book at a recent seminar hosted by APARC's Taiwan Program. Watch the session recording:

Sign up for APARC newsletters to receive event invitations >

Challenging the stereotype of "The West innovates, and the East imitates," Tinn traces the development of Taiwan’s computer and semiconductor industry to efforts by Taiwanese hobbyists and enthusiasts who leveraged Cold War-era U.S. technical assistance while simultaneously adapting and modifying imported technology. Through such hands-on tinkering, they deepened their technical understanding, made computing accessible across the island, and built the foundation for domestic manufacturing.

In her seminar presentation, Tinn focused on Chapter 8 of her book, which describes the technological achievements of Acer's founder, Stan Shih, and his engineers, and critiques their portrayal as counterfeiters by U.S. stakeholders. She recounts how, in 1983, Shih's company, originally named Multitech, realized the dream of manufacturing computers domestically and shipped a thousand microcomputers to the United States, only to have them intercepted by U.S. Customs in San Francisco. The computers, called Micro-Professor II, were deemed Apple II counterfeits despite being independently designed with unique features. Although Apple did not file suit against Multitech in Taiwan, it worked intensely to prevent Multitech from exporting its products to the United States.

What followed, Tinn argues, revealed deep misunderstandings, sometimes deliberate, about Taiwan's technological capabilities. U.S. media and analysts failed to make sense of Shih’s success in the microcomputer market, and Congressional hearings in 1983 portrayed Taiwanese computer manufacturers as counterfeiters and invaders threatening American industry.

This misrepresentation, she says, reflected Orientalist assumptions that Taiwanese lacked the capacity to innovate. Tinn writes:

"The Orientalist representations of Taiwanese computer manufacturers can be selfcontradictory. On the one hand, when it comes to the manufacturing capacity of Taiwanese computer makers, participants in the congressional hearings exaggerated the numbers of counterfeit computers Taiwanese companies dumped in the United States. On the other hand, more and more US computer manufacturers gradually recognized the manufacturing capacity of Taiwanese companies and subcontracted them to make computers at lower cost for the expanding US and global personal computer market."

Island Tinkerers is available from MIT Press, including in an open-access edition. 

The deliberations yielded clear priorities for building trust through safeguards during this early phase of agentic development and adoption. Currently, participants favor AI agents for low-risk tasks, while expressing caution about high-stakes applications in medical or financial domains. In deliberation, participants indicated an openness to these higher-risk applications if provided safeguards around privacy or user control, such as requiring approval before finalizing an action.

The Forum also revealed support for culturally adaptive agents, with a preference for asking users about norms rather than making assumptions. Lastly, the discussions underscored the need for better public understanding of AI agents and their capabilities, pointing to the importance of transparency and education in fostering trust in these emerging technologies.

"The perspectives coming out of these initial deliberations underscore the importance of our key focus areas at Cohere: security, privacy, and safeguards,” said Joelle Pineau, Chief AI Officer at Cohere. “We look forward to continuing our work alongside other leaders to strengthen industry standards for this technology, particularly for enterprise agentic AI that works with sensitive data."

This pioneering forum sets a new standard for public participation in AI development. By seeking feedback directly from the public, combining expert knowledge, meaningful public dialogue, and cross-industry commitment, the Industry Wide Forum provides a key mechanism for ensuring that AI innovation is aligned with public values and expectations.

“Technology better serves people when it's grounded in their feedback and expectations,” said Rob Sherman, Meta’s Vice President, AI Policy & Deputy Chief Privacy Officer.  “This Forum reinforces how companies and researchers can collaborate to make sure AI agents are built to be responsive to the diverse needs of people who use them – not just at one company, but across the industry.”

Through Stanford’s established methodology and their facilitation of industry partners, the Industry-Wide Forum provides the public with the opportunity to engage deeply with complex technological issues and for AI companies to benefit from considered public perspectives in developing products that are responsive to public opinion. We hope this is the first step towards more collaboration among industry, academia, and the public to shape the future of AI in ways that benefit everyone.

“We have more industry partners joining our next forum later this year”, says Alice Siu, Associate Director of Stanford's Deliberative Democracy Lab. “The 2026 Industry-Wide Forum expands our discussion scope and further deepens our understanding of public attitudes towards AI agents. These deliberations will help ensure AI development remains aligned with societal values and expectations.”