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.”

Image

At a recent seminar hosted by APARCʼs China Program, Professor Jessica Chen Weiss, the David M. Lampton Professor of China Studies at the Johns Hopkins University's School of Advanced International Studies, presented findings from her forthcoming book, Faultlines: The Tensions Beneath China's Global Ambitions (under contract with Oxford University Press), which examines how domestic politics and regime insecurity shape China’s foreign policy ambitions, prospects for peaceful coexistence, and the future of international order. Drawing on research and fieldwork in China, Weiss argued that understanding Beijingʼs behavior on the world stage requires looking beyond ideology to the contested priorities and political calculations that drive decision-making within the Chinese Communist Party (CCP).

Weiss proposed a framework centered on three pillars that have sustained CCP legitimacy since the late 1970s: sovereignty (nationalism), security (civility), and development. Her analysis reveals that China's objectives are not static but moving targets shaped by competing domestic interests, leadership priorities, and international pressures.

Sign up for APARC newsletters to receive our event invitations and programming highlights >

The Sovereignty-Security-Development Paradigm
 

At the heart of Weissʼs argument is the recognition that the CCPʼs foremost concern is domestic survival. In the face of the collapse of most communist regimes, the Party has remained vigilant against what it calls “peaceful evolution” and democratic contagion.

On issues touching core sovereignty concerns – Taiwan, Hong Kong, and maritime territorial claims – China has been “hyperactive” in making demands, even when doing so invites international censure. Weiss explained that the more central an issue is to CCP domestic legitimacy, the harder it becomes to make concessions, and the more likely international pressure is to backfire.

Yet tensions exist between competing priorities. China has compromised on certain border disputes to shore up domestic security, while its evolving stance on climate change reflects a shift from viewing carbon limits as threats to growth to recognizing the greater threat environmental catastrophes present to the nation’s stability.

Beyond the Monolith: China's Internal Contestation
 

Weissʼs research demonstrates that authoritarian China is far from monolithic. Different geographic, economic, institutional, and even ideological interests shape policy debates, even if most actors lack formal veto power. Local governments can resist central directives, as evidenced during the COVID-19 outbreak, when local officials initially withheld information about human-to-human transmission from the central government to prevent panic from disrupting important political meetings.

This pattern of center-local tension extends to China's international commitments. Local officials often game environmental regulations to juice growth and secure promotions, undermining Beijingʼs pledges on carbon emissions. On issues ranging from Belt and Road investments to export controls, implementation frequently diverges from stated policy as local actors pursue their own interests.

Weiss’s framework distinguishes among issues that are both central and uncontested (such as Taiwan), those that are central but contested (like climate change and trade policy), and peripheral issues where Beijing has shown greater flexibility (such as demonstrated by many UN peacekeeping initiatives). This helps explain why international pressure succeeds in some domains but fails spectacularly in others.

"The more central an issue is domestically, the more pressure the government faces to perform, and the harder it is to defy these domestic expectations," Weiss said. As a result, international pressure on these central issues is more likely to backfire, forcing the government to be seen as defending its core interests. She underscored that "even on these central issues, there's often tension with other central priorities, and managing these trade-offs comes with a number of different risks. It also means that sometimes an issue that touches on one pillar of regime support can yield to another."

Nationalism as Constraint and Tool
 

Weiss described nationalism as both a pillar of the CCPʼs legitimacy and a potential vulnerability when the government’s response appears weak. While large-scale anti-foreign protests have become rare, nationalist sentiment remains active online and shapes diplomatic calculations.

During Speaker Nancy Pelosiʼs 2022 visit to Taiwan, Chinese social media erupted with calls for the PLA to shoot down her plane. One interlocutor told Weiss his 14-year-old son and friends had stayed up past bedtime to watch Pelosiʼs plane land, illustrating nationalismʼs penetration into Chinese society.

Survey research reveals Chinese public opinion is quite hawkish, with majorities supporting military spending and viewing the U.S. presence in Asia as a threat. The government often refrains from suppressing nationalist sentiment to avoid backlash, even when doing so creates diplomatic complications. Weissʼs public opinion survey experiments, however, reveal that tough but vague threats can provide the government with wiggle room for de-escalation, although disapproval emerges when action is not sufficiently tough.

Regime Security Without Ideological Crusade
 

Weiss pushed back against arguments that China is bent on global domination or that ideology drives conflict with the West. While the CCP seeks a less ideologically threatening environment, it must balance this against development and market access.

This pragmatic calculus explains China's constrained support for Russiaʼs war in Ukraine — Beijing fears secondary sanctions more than it values autocratic solidarity. Rather than exporting revolution, China has worked with incumbents of all political stripes in the service of its national interests.

Chinaʼs strategy focuses on making autocracy viable at home, not on defeating democracy globally. This suggests room for coexistence if both sides can reach a détente on interference in internal affairs.

“China's activities are making autocracy more viable and, to the extent that China is succeeding, making China's example more appealing as a consequence. But its strategy doesn't hinge on defeating democracy around the world,” argued Weiss. This implies, to her view, that “there is more room for coexistence between autocracies and democracies if these different systems can find or reach a potential détente in the realm of ideas about how countries govern themselves, and importantly, they need to pull back their efforts in other societies across boundaries.”

Interdependence and Future Trajectories
 

Weiss concluded by outlining how her framework suggests different engagement strategies depending on where issues fall within the centrality-contestation matrix. On central but uncontested issues like Taiwan, pressure proves counterproductive, and reciprocal restraint may be most promising. On central but contested issues like currency, multilateral pressure can influence certain Chinese constituencies against others. On peripheral issues, such pressure is most effective unless powerful domestic constituencies subvert implementation.

Addressing questions about U.S.-China decoupling, Weiss noted that both sides recognize there are interdependencies that don’t have quick solutions. Even in a critical area like minerals, diversification will take at least a decade, and Chinese processing will still dominate globally. The goal of diversification should be to preempt coercion, not to achieve true decoupling.

Introduction and Motivation:

Social scientists and philosophers have shown increased interest in the concept of descriptive representation. Indeed, the identities and life experiences of those doing the representing may be critical for enacting the preferences of those being represented. This is especially the case for historically marginalized groups — upper-class, educated men may simply fail to adequately represent lower-class, uneducated, or women voters. In addition to considerations of fairness, there is growing recognition that descriptive representation can improve policy outcomes, such as service delivery or trust in government.

However, the study of descriptive representation has been hampered by data availability. Identifying simple correlations across the world’s democracies — for example, between proportions of working-class legislators and levels of social welfare provision — has hitherto been impossible.

In “The Global Legislators Database,” Nicholas Carnes, Joshua Ferrer, Miriam Golden, Esme Lillywhite, Noam Lupu, and Eugenia Nazrullaeva introduce the largest dataset of biographic and demographic information on national legislators ever assembled. GLD will enable scholars to assess just how much voters elect those with life experiences resembling their own. The authors compile information on five descriptive variables across 97 democracies: legislators’ party affiliation, gender, age, highest level of education, and previous occupation (to assess their social class). By contrast, prior datasets have focused only on heads of state or cabinet members, or on only a selection of more developed democracies.

Some questions around descriptive representation would seem to have intuitive answers: Wouldn’t developed countries have more women representatives, or wouldn’t women legislators feature less prominently in right-wing parties? Scholars can now hope to do more than merely gesture at answers.

Characteristics, Validity Checks, and Applications:

GLD comprises countries that (a) have a population of over 300,000 and (b) meet the standard for what Freedom House calls “Electoral Democracy” — having some minimum of political rights and civil liberties. Excluding six cases of data availability constraints, this yields 97 countries, including India, the United States, Brazil, Mexico, Nigeria, and the Philippines. Scholars of Pakistan, Bangladesh, or Turkey — who would likely characterize these countries as authoritarian during the 2015-17 time period — will be pleased that the authors chose a more forgiving measure of democracy.

Biographic data is drawn from the national legislature in unicameral countries and the lower chamber in bicameral countries. (By contrast, upper chambers are sometimes indirectly appointed or hereditary, which sheds less light on whether voters choose descriptively similar representatives.) This yields over 19,000 individuals who held office during at least one legislative session in 2015, 2016, and 2017. GLD has remarkably complete data for the five variables mentioned above: age and education data are presented for over 90% of legislators in the dataset, for over 93% as regards occupation, and nearly 100% for gender.

In order to assess GLD’s validity, the authors compare select variables to those in comparable datasets. For example, the gender variable is compared with gender data from the Varieties of Democracy (V-Dem) project, which shows that the two measures are nearly identical. So too is the age data nearly identical to an index from 15 affluent democracies.
 

Image

Figure 1. Shares of women legislators in the GLD and V-Dem. Note: Bahamas, Belize, Fiji, and Kosovo are omitted because of missing data in the V-Dem.
 

For categories like education and prior occupation — where comparable data are unavailable — the authors conduct “face validity” tests: these draw on our intuitions that legislators are, for example, mostly educated and not working class. And indeed, these intuitions are borne out in the distributions of GLD data. In terms of total coverage, GLD includes information on more legislators than the comparable Global Leadership Project database for all but two countries, and in many cases, the differences are large.
 

Image

Figure 3. Distributions of legislator traits in the GLD. Note: Age is calculated at the time of election. Higher education includes levels beyond primary and secondary education (Bachelors, Masters, PhD, LLB, LLM, JD, MD, and short-cycle tertiary). Data on educational attainments for legislators is unavailable for Côte d’Ivoire.
 

The authors then use GLD in application to a number of questions for which scholars have lacked global data. First, some have hypothesized that in legislatures with more (a) uneducated, (b) female, and (c) working-class representatives, incumbency rates will be lower. This is because individuals from these three groups might have a harder time overcoming challenges relating to expertise, sexism, and fundraising, respectively. Correlating GLD data with a global reelection database, the authors find only evidence for (b), suggesting that women may face higher barriers to remaining in office. These are only correlations, but they point to fruitful areas for exploration: why might women face unique barriers, and what distinguishes countries with lower versus higher barriers?
 

Image

Figure 5. Re-election rates by years of education, gender, and occupational background. Note: The share of working-class legislators is zero for six countries that are dropped from the figure: Albania, Botswana, Cyprus, Estonia, Guatemala, and Mongolia.
 

A second application involves public financing of elections, which is thought to favor more working-class legislators who would have a harder time fundraising. Correlating V-Dem data on public financing with the GLD variable on prior occupation, however, the authors find limited evidence for this conjecture. Finally, some have proposed that countries with a stronger rule of law would favor a higher proportion of lawyers in the national legislature. Looking again at GLD prior occupation data alongside V-Dem rule of law data, the authors find limited evidence for this hypothesis.

These varied applications point to how the Global Legislators Database can serve as a valuable resource for scholars interested in the causes and consequences of descriptive representation. Although the GLD covers only a single point in time, it can serve as a bedrock for additional data-collection efforts. In addition to expanding its temporal coverage, scholars may also wish to gather data on upper chambers. Especially in ethnically diverse countries like Bosnia and Herzegovina, upper chambers are intended to mirror the descriptive composition of specific regions. However, it may be the case that ethnic representatives are still vastly more wealthy or educated than their constituents, thus impeding their ability to represent.

*Brief prepared by Adam Fefer.