Motivation & Contribution

Over the past 10-15 years, both longstanding and relatively new democracies have suffered from backsliding and erosion, including India, the United States, Brazil, Turkey, and many others. Many social scientists have explained this wave of backsliding in terms of either (a) elected autocrats who undermine democracy from within or (b) declining popular support for democrats who have failed to deliver economic growth and prosperity. However, recent scholarship by Thomas Carothers and Brendan Hartnett has questioned the wisdom of the latter. For example, India enjoyed strong economic growth prior to its backsliding under Narendra Modi.

In “Delivering for Democracy,” Francis Fukuyama, Chris Dann, and Beatriz Magaloni set out to more systematically evaluate the evidence connecting popular support for democracy with delivery, examining both backsliding and non-backsliding countries. After finding preliminary evidence for the democracy-delivery relationship, they offer an explanation of why delivery is simultaneously so important and so elusive under democratic governance.

Evidence

Using ten data sources covering 650,000 people in both old and new democracies, the authors find a strong, positive correlation between satisfaction with democracy and economic performance. This relationship holds not only for many countries at one point in time but for pairs of countries over time. In two developing democracies — Argentina and Brazil — as well as in two developed democracies — Greece and Spain — satisfaction and delivery have been closely connected since 2005, plummeting during economic crises and rising during periods of prosperity. These patterns call for an explanation for why voters care so much about delivery, such that they may be willing to sacrifice their democratic freedoms for it.
 

Image

Image

The Argument

Delivery is Important

The authors begin from the axiom that stable political life depends upon citizens perceiving their governments as legitimate. Legitimacy can be thought of in terms of both performance — the effective delivery of goods and services — and procedure, which encompasses policies that reflect the democratic will of the people. As the examples of China, Turkey, Saudi Arabia, and Singapore show, however, plenty of autocracies and backsliding democracies have not only delivered, but have also arguably outperformed their democratic peers. From China’s Belt and Road Initiative to Turkey’s Yavuz Sultan Selim Bridge, authoritarian leaders and ruling parties have achieved remarkable performance legitimacy. 

Although autocracies, by definition, cannot be procedurally legitimate, this may carry little weight for democratic citizens who experience prolonged unemployment or must deal with dilapidated infrastructure. Indeed, public engagement through procedural channels — such as voting or jury service — has steadily declined across the democratic world. Democratic voters are increasingly willing to support outsider candidates who build new infrastructure or promise to fight crime, but who nonetheless restrict their political freedoms. Many citizens of El Salvador — which now claims the world’s highest incarceration rate — continue to view Nayib Bukele’s administration as the surest way of delivering security, despite a years-long state of emergency that has seriously eroded democratic freedoms. 

Meanwhile, established democracies built much of their infrastructure decades ago, so investments primarily maintain these systems, rather than showcasing new projects that can garner public support. In some cases, democracies have struggled to even maintain their existing infrastructure, perhaps best exemplified by the collapse of Baltimore’s Francis Scott Key Bridge. All of this creates the conditions for voters to support far-right, anti-democratic parties, which often blame immigrants for economic problems and propose illiberal solutions.

Yet Democratic Delivery is Difficult

Elected democrats seeking to deliver may find themselves hamstrung in ways that autocrats are not. For one, democratic institutions are composed of ‘veto players’ who can stymie the introduction of badly needed policies: national and subnational governments, multiple legislative chambers, judges who review and overturn executive action, and so on. At the same time, democrats must worry about election cycles and term limits, decreasing their incentives to deliver for the long term when later politicians may take credit. Meanwhile, legal and regulatory systems, such as those intended to protect the environment, may prevent the building of critical infrastructure. Property rights prevent the forcible displacement of communities for development, while civil liberties prevent the repression of those who refuse to be displaced. Rules meant to prevent regulatory capture often become arenas where powerful interest groups block and delay government action. 

Independent news media present another potential impediment to delivery, as criticism from journalists can make incumbents wary of undertaking new projects. In addition, media bias can convince voters to remove politicians who do, in fact, deliver. By contrast, autocrats who censor media and arrest journalists can focus on delivery alone, even while their development schemes often rest on corrupt and nepotistic practices. Popular discontent with democratic government ultimately creates a damaging feedback loop: voters are unwilling to fund government projects, in turn leading government to function worse, generating further discontent.
 

Prospects

Autocrats have figured out ways to deliver the goods and services their citizens want, but this does not make autocracy a just political system. By the same token, democracies may struggle to deliver, but their procedural legitimacy — especially voters’ ability to hold representatives to account — entails a powerful means of generating fair and inclusive delivery. As such, the authors call on democracies to examine their past and that of their peer countries — both developed and developing — for inspiration. For example, the U.S. New Deal was exemplary in building ambitious and popular infrastructure, as well as providing broad social and economic protections. (Of course, most of these projects would be hamstrung by modern-day regulatory frameworks.)

Meanwhile, Australia’s citizens have both benefited from a recent infrastructure boom and have demonstrated strong support for democracy. Finally, many Latin American countries have implemented popular and effective programs like conditional cash transfers. For the authors, addressing the issues most pressing to voters — such as job creation, which is especially salient to young people, who are most dissatisfied with democracy — will require democratic governments to strike the right balance between democracy and delivery.

*Research-in-Brief prepared by Adam Fefer.

Motivation & Contribution

China has become a world leader in digital repression, including social media surveillance, internet shutdowns, website-blocking, and arrests of those posting content the Chinese Communist Party (CCP) deems threatening. These technologies have been deployed not only on Chinese citizens but have also been exported around the world, to both democracies and dictatorships. Huawei, the world’s largest — and partly state-owned — telecommunications provider, is at the center of this dynamic, having been exported to over 60 countries on every continent.

Despite the visibility of Huawei technology transfers outside of China, our understanding of how exactly this affects recipient countries is still in its infancy. Some observers see Huawei as a cornerstone of the CCP’s efforts to prop up foreign dictatorships. Others see Huawei as a source of economic growth in some of the world’s most populous and least developed countries, providing cost-effective telecommunications and infrastructure development. What is needed is systematic evidence to move beyond speculation and to better understand whether and how Huawei transfers affect repression.
 

Image

Erin B. Carter and Brett L. Carter help fill this crucial gap in our knowledge. Using data on all Huawei contracts across the world over a nearly twenty-year period, they find that Huawei transfers do, in fact, facilitate digital repression in the autocracies — but not the democracies — that receive them. Attempts at digital repression in democracies are more often checked by independent political institutions and legal frameworks. When these fail, civil society organizations, news media, and popular mobilization can step in. By contrast, autocrats tend to lack these constraints and can thus more readily use Huawei for digital repression. In short, the impact of Huawei transfers varies depending on regime type.

The article compellingly demonstrates the enduring role of political institutions — in this case, those enabling democratic resilience. In addition, it sheds light on one mechanism underlying the so-called global wave of autocratization, namely, Huawei transfers that enable dictators to more efficiently control their citizens. And from a methodological standpoint, the paper provides the “first cross-country, plausibly causal evidence” of the effects of Huawei on repression.

Data & Methods

The authors use data on Huawei contracts from 2000 to 2017, which comes from the AidData project. This includes 153 projects — almost entirely on communications technologies — in 64 countries worth over $1.5 billion. The vast majority of recipient countries are in Sub-Saharan Africa and Central, South, and Southeast Asia. Huawei is responsible for around 70% of all telecommunications in Africa alone. The single largest recipient of Huawei transfers is Indonesia (~$391 million of contracts), followed by Uzbekistan (~$150 million) and Cameroon (~$100 million). Notably, of the top 20 recipient countries, only one (Costa Rica) is considered “Free” by Freedom House, the rest being “Partly Free” or “Not Free.” Huawei transfers flow slightly more to democracies than to autocracies, but this is only if Indonesia is included; otherwise, transfers flow disproportionately to autocracies.
 

Image

Carter and Carter quantify the effect of Huawei transfers on repression by using the generalized synthetic control (GSC) method. The goal is to compare those countries receiving (the “treatment”) of Huawei transfers with those in the “control group” that do not. However, comparing the treated and controlled countries or “units” is difficult for several reasons. For one, the countries that receive Huawei transfers differ from those that do not, for example, in terms of their political institutions and economic development. In addition, there is a lag of several years between receipt of Huawei and its full implementation, and the effects of Huawei transfers may take several years before they are fully realized. Similarly, transfers are staggered over time; for example, the largest number of contracts was signed as China’s Belt and Road Initiative expanded between 2014 and 2015.

Comparing (such ostensibly different) treated and controlled units can be facilitated by constructing a “synthetic” control out of multiple control units. This will more closely match the treated unit’s characteristics prior to receiving Huawei. GSC extends (or generalizes) the synthetic control method by enabling comparison across multiple time periods (given the staggered rollout mentioned above), and across countries with different backgrounds prior to adopting Huawei technologies. Separate tests are then run for autocracies and democracies to capture how political institutions influence the effects of Huawei transfers.

Findings

In addition to the central finding about regime type, the authors also provide evidence that the countries most likely to contract with Huawei are more populous and poorer. These countries are, from China’s standpoint, attractive markets, and their governments appreciate the subsidies offered by China’s Export-Import Bank. As such, transfers are driven, in part, by demand in recipient countries. In addition, recipient countries are also more likely to have received Chinese aid and, therefore, recognize the One China policy. Meanwhile, autocracies, coup-prone countries, and those rich in natural resources are not more likely to receive Huawei transfers. All of these findings should be qualified by the secrecy surrounding Huawei contracts. Indeed, many such contracts include confidentiality clauses that prohibit recipient countries from divulging information.
 

Image

Correlates of Huawei technology transfers
 

Another set of findings relates to those democracies that do use Huawei to digitally repress, even if transfers do not systematically increase repression in democracies. India is an important case, as it is a world leader in periodically shutting down the internet. Unsurprisingly, democracies whose political institutions and civil society organizations have been weakened by elected leaders with authoritarian ambitions (such as Narendra Modi) seem more likely to impose shutdowns or monitor media. At the same time, weak democracies are unlikely to use Huawei to block websites or arrest their citizens for digital action. As Carter and Carter point out, this highlights the importance of disaggregating the concept of digital repression as well as the set of institutions that enable democratic resilience. Ultimately, “Exporting the Tools of Dictatorship” is exemplary in its use of systematic data to shed light on a complex and contested geopolitical issue.

*Research-in-Brief prepared by Adam Fefer.

Motivation & Contribution

Vladimir Putin has ruled Russia for over a quarter century despite economic mismanagement, corruption, and a disastrous war in Ukraine. Prior to the invasion, Putin’s rule was often seen as exemplifying modern autocracy, maintaining one-man rule while employing less overt mechanisms of social control, such as propaganda and regularly held, albeit controlled, uncompetitive elections. Since the full-scale invasion of Ukraine in February of 2022, however, Putin’s authoritarianism has intensified, especially as opponents of the war are imprisoned, terrorized, driven into exile, or worse. As such, Russia’s “modern” autocracy, which relies less on force than on presenting a message to society of an all-capable state and leader, has transformed into a highly personalized form of harsh authoritarianism. At the same time, many average Russians do genuinely support Putin — especially given his emphasis on order in society and restoring Russian greatness — and view the regime as legitimate and responsive. Often, however, analysts of contemporary Russia in the West tend to focus more on those Russians who oppose Putin’s autocracy than on those who support it and why they do. 

In her “Review Article: Why Does Putinism Endure? Dialoguing With the Dictator,” Kathryn Stoner tackles this question through a critical analysis mainly of Hannah Chapman’s Dialogue with the Dictator: Authoritarian Legitimation and Information Management in Putin’s Russia. The analysis is complemented by insights from four other recently published books on authoritarianism and Russia. For Stoner, Chapman’s central contribution lies in showing how a seemingly ridiculous exercise in state propaganda — the Direct Line with Vladimir Putin TV call-in show — is in fact an important way of legitimizing Putin’s rule, providing his regime with information about social preferences and sources of dissatisfaction while affording citizens with the feeling that their complaints are being heard and even resolved live and on TV. Gaining information about citizens would otherwise be difficult in such an unfree society, where individual preferences that may be critical of Putin’s rule are often concealed or deliberately falsified to avoid punishment by the regime. The reader learns that even hegemonic dictators must be perceived by citizens to be a good listener. 

The Functions of ‘Participatory Technologies’

The conceptual foundation of Chapman’s analysis lies in what she terms ‘participatory technologies’ (PTs), which allow citizens to interact with authoritarian regimes “directly and in a positive manner.” Observers tend to dismiss PTs, which are often low-technology formats like TV shows or radio programs, as ‘spin’ or mere propaganda. In Russia, however, Putin uses PTs not only to control what information reaches citizens, but also to enhance his legitimacy and appear more accountable. Although PTs are limited and highly individualized — for example, more often than not ignoring collective demands — they are direct.

Participatory technologies differ from autocrats’ sham, window-dressing elections, where the outcome is a foregone conclusion. Instead, they permit citizens to engage with and provide input to the autocrat on specific issues. They also provide information to the regime about the performance of lower-level agents, such as governors or regional politicians, given that citizens are often concerned with local, day-to-day problems. By involving these lower-level agents, autocrats can deflect responsibility for governance problems away from themselves. Finally, PTs provide value to those who participate in them, giving them a sense of political efficacy. At the same time, however, PTs may expose embarrassing governmental dysfunctions. Indeed, the resolution of local problems – like fixing potholes on a local road – should not require direct citizen engagement with powerful national-level leaders. Furthermore, PTs run the risk of being overused and being seen as performative and fake by citizens.
 

Tuning into Direct Line

Stoner’s review article discusses Chapman’s analysis of one PT in particular, namely the Direct Line with Vladimir Putin call-in show. Direct Line began in 2001, organized by the Kremlin and broadcast on state-owned and pro-regime media. The audience and the questions posed are selected in advance. During the broadcast — which is one of the most watched events in Russia, perhaps because of compulsion from employers — Putin conveys to the audience that Russia is doing well, but that he is not satisfied, is working tirelessly, and wants to hear citizens’ concerns. Putin does not merely listen but acts to solve their concerns, for example, keeping a folder of issues raised by ordinary Russians and bringing these issues to the attention of regional governors. He has even forwarded questions directly to regional governors on air, which also serves to deflect responsibility away from himself. 

Methodology

Stoner describes the multiple survey experiments and interviews Chapman conducted in exploring whether Direct Line matters to ordinary Russians. Some of the interviewees seem to know that Direct Line is staged; yet they nonetheless find it compelling that Putin is not afraid to answer their sometimes uncomfortable questions. The goal of the survey experiments is to try to ascertain how prompts that (do not) mention Direct Line affect respondents’ evaluations of Putin. In other words, Direct Line is the ‘treatment,’ which participants in the ‘control’ group do not receive. One experiment only varies whether or not Putin heard about a problem on Direct Line. Another varies this as well as whether or not Putin resolved the problem. In both experiments, respondents whose prompts included mention of Direct Line evaluated Putin more favorably, although Stoner notes that the statistical effects of this are small. In another experiment, the mention of Direct Line leads respondents to believe there are greater opportunities to express their opinions and demands. Finally, a fourth experiment finds that respondents who support (oppose) Putin gave him higher (lower) evaluations when Direct Line was mentioned, hinting at how Direct Line might serve as a source of polarization in Russia.
 

Limitations

Stoner highlights an important puzzle about whether participatory technologies like Direct Line can, in fact, provide credible information to Putin’s regime. For example, questions are pre-screened to exclude subjects that could threaten the regime. Additionally, those who call in are already likely to be more supportive of or sympathetic to the regime. It would thus seem that Putin has better mechanisms for gathering credible information, such as through the Kremlin’s surveillance system, than through a PT like Direct Line, as Chapman claims. As such, Direct Line’s value appears primarily rhetorical, as opposed to informational — the regime is merely performing an exercise in responding to easy problems that can be resolved quickly by Putin, or inane questions about things like Putin’s favorite books.

Stoner’s review article notes her concerns regarding the generalizability of Chapman’s findings. First of all, it would be challenging to draw conclusions from the survey experiments, given that they concern a single group of people who were surveyed a decade ago (2015-2016). Relatedly, and in light of the war in Ukraine, Chapman’s analysis of PTs seems dated, as Putin has resorted far more often to severely punishing opposition (real and imagined) to his regime, rather than cajoling and coercing society with participatory technologies like his call-in show. Indeed, the show was cancelled in 2022 without explanation, but presumably to avoid difficult questions about the devastation wrought that autumn on the Russian armed forces in Ukraine. Nonetheless, Stoner notes that there is now a burgeoning literature that helps us better understand how, in normal times, autocrats effectively utilize the media and other technologies to build support and legitimate their rule, while providing citizens with a sense that they can influence politics, when in reality, they cannot.

Argument & Key Findings

The system of reentry institutions — including halfway houses, parole agencies, and housing assistance programs — can be extremely complicated for formerly incarcerated persons (FIPs) to navigate. These complications are not merely logistical, but also social and emotional: the ways in which FIPs interact with reentry institutions can affect their sense of belonging, dignity, and prosperity. When the rules and practices of reentry institutions undermine these needs, it becomes more likely that participants will violate rules, withdraw from the institutions altogether, or find themselves reincarcerated. 

In “Home But Not Free,” Gillian Slee offers a rich analysis of these socioemotional dynamics. The paper both increases our knowledge of reentry processes and deepens our understanding of FIPs and reentry staff. Previous scholars have focused more on how adverse outcomes stem from reentry institutions prioritizing surveillance or control. Slee pushes forward this conversation by highlighting how adverse outcomes also stem from failures to acknowledge and support the dignity of FIPs.

Slee’s paper is informed by over two years of ethnographic fieldwork with FIPs in Philadelphia. This includes observing over 200 appointments at a housing assistance program, analyzing more than 130 files of program participants, and both observing and assisting with programming at a women’s halfway house. 
 

Three Mechanisms:

The core of the paper centers on Slee’s elaboration of three mechanisms that link socioemotional concepts, such as indignity, to outcomes like withdrawal or reincarceration. Each mechanism is clarified through a range of examples and case studies.

1. Unrecognized vulnerability:

Reentry institutions and their staff often fail to recognize or respond to the constraints and vulnerabilities faced by FIPs. These failures can undermine the dignity of FIPs and provoke their withdrawal from such institutions. In the realm of housing assistance, the Philadelphia program requires that rent falls between 30% and 45% of participants’ income. However, this is often unrealistic given the difficulties of finding fairly priced units or securing gainful employment with a felony conviction. Accordingly, FIPs must seek low-quality units or roommates. Yet most participants do not want roommates because it reminds them of being incarcerated. Participants are thus presented with an undignified set of choices. The program restrictions mean that many FIPs cannot or will not utilize housing assistance programs, deepening their sense of instability.

Another source of vulnerability concerns the mismatch between FIPs’ expectations and the realities of frontline bureaucracy. For example, many housing assistance programs have too few staff, some of whom struggle to juggle appointments or return phone calls. Because reentry staff are overburdened, they may ask participants to pick up the slack by searching for housing units. Yet many FIPs lack the requisite know-how, for example, calling about units too frequently or too early in the morning. Others may show up hours early for their appointments, in the process annoying reentry staff. Yet participants are not coached on how to improve these behaviors, leading to neglect. Other FIPs must learn that the majority of units are listed online as opposed to in newspapers, incurring the mockery from reentry staff in the process. 

A final source of vulnerability concerns participants’ lack of efficacy — the sense that their efforts make little difference or are inadequate. Reentry staff may have high expectations of people who feel “cryogenically frozen in time” (p. 32) because of years of incarceration. Some are unable to use modern cell phones or have no rental history.
 

2. Discretion’s Benefits and Drawbacks:

The discretion exercised by reentry staff introduces difficult choices for participants, forcing them to choose between (a) following the rules and becoming socially isolated or (b) breaking the rules and developing social connections. For example, some halfway houses are restrictive about time spent outside of the house. Participants who abide by the rules may miss out on socially important events, like a child’s basketball game. Some FIPs may lie about or conceal where they live in order to deal with less intrusive parole agents. Others may cross state lines to pursue important career opportunities. One participant parked their mobile home outside of the parole district lines because it was less expensive and easier than seeking alternative units, but these kinds of ‘rational’ behaviors cannot be accommodated. Discretion is a highly variable attribute: some reentry staff cancel meetings and inconvenience participants, while others remember individuals’ needs and accommodate them. Those who expect more discretion than they receive may break the rules out of frustration. Ultimately, discretion and its absence can provoke a host of socioemotional problems.
 

3. Risk-Escalating Rules:

Instead of preventing noncompliance, program rules may serve to encourage it when they undermine participants’ sense of dignity. For example, 29 states prohibit associating with other FIPs, yet many participants have friends or family supervised by the system; as such, people violate the rules in order to preserve meaningful relationships. Some FIPs are faced with painful dilemmas, for example, choosing between living in halfway houses where drug use is common or breaking the rules by leaving. Others report using cocaine instead of marijuana because the latter can be detected in their bloodstream for much longer. Some halfway houses mandate spending a certain number of hours inside the house, but this leads to participants being unable to work multiple jobs to support themselves, a clear violation of their dignity.

For many FIPs in uncomfortable halfway houses, they cannot be placed in another house unless they break the rules of their existing house; some consider breaking the rules for the sake of their well-being, even though doing so might land them under even more restrictive supervision. One participant was refused permission to live in a camper that he could afford because the camper’s mobility posed a flight risk. Another participant broke the rules by traveling out of state because her son’s father had cancer, and a reentry professional later told her to return in a rather threatening way. The rules of reentry institutions thus incentivized FIPs to make very risky choices.
 

By highlighting socioemotional concepts — especially (in)dignity — as central to the experiences of formerly incarcerated persons, Slee shows how the rules and practices of reentry institutions can undermine reintegration. Addressing sources of vulnerability and counterproductive rules may help reform reentry institutions in more humane and effective ways.

Introduction

Generative AI has become an incredibly attractive and widespread tool for people across the world. Alongside its rapid growth, AI tools present a host of ethical challenges relating to consent, security, and privacy, among others. As Generative AI has been spearheaded primarily by large technology companies, these ethical challenges — especially as viewed from the vantage point of ordinary people — risk being overlooked for the sake of market competition and profit. What is needed, therefore, is a deeper understanding of and attention to how ordinary people perceive AI, including its costs and benefits.

The Meta Community Forum Results Analysis, authored by Samuel Chang, James S. Fishkin, Ricky Hernandez Marquez, Ayushi Kadakia, Alice Siu, and Robert Taylor, aims to address some of these challenges. A partnership between CDDRL’s Deliberative Democracy Lab and Meta, the forum enables participants to learn about and collectively reflect on AI. The impulse behind deliberative democracy is straightforward: people affected by some policy or program should have the right to communicate about its contents and to understand the reasons for its adoption. As Generative AI and the companies that produce it become increasingly powerful, democratic input becomes even more essential to ensure their accountability. 

Motivation & Takeaways

In October 2024, the third Meta Community Forum took place. Its importance derives from the advancements in Generative AI since October 2023, when the last round of deliberations was held. One such advancement is the move beyond AI chatbots to AI agents, which can solve more complex tasks and adapt in real-time to improve responses. A second advancement is that AI has become multimodal, moving beyond the generation of text and into images, video, and audio. These advancements raise new questions and challenges. As such, the third forum provided participants with the opportunity to deliberate on a range of policy proposals, organized around two key themes: how AI agents should interact with users and how they should provide proactive and personalized experiences for them.

To summarize some of the forum’s core findings: the majority of participants value transparency and consent in their interactions with AI agents as well as the security and privacy of their data. In turn, they are less comfortable with agents autonomously completing tasks if this is not transparent to them. Participants have a positive outlook on AI agents but want to have control over their interactions. Regarding the deliberations themselves, participants rated the forum highly and felt that it exposed them to alternative perspectives. The deliberators wanted to learn more about AI for themselves, which was evidenced by their increased use of these tools after the deliberations. Future reports will explore the reasoning and arguments that they used while deliberating.
 

Image

^{The participants of this Community Forum were representative samples of the general population from five countries - Turkey, Saudi Arabia, India, Nigeria, and South Africa. Participants from each country deliberated separately in English, Hindi, Turkish, or Arabic.}

Methodology & Data

The deliberations involved around 900 participants from five countries: India, Nigeria, Saudi Arabia, South Africa, and Turkey. Participants varied in terms of age, gender, education, and urbanicity. Because the deliberative groups were recruited independently, the forum can be seen as five independent deliberations. Deliberations alternated between small group discussions and ‘plenary sessions,’ where experts answered questions drawn from the small groups. There were around 1000 participants in the control group, who did pre- and post-surveys, but without deliberating. The participant sample was representative with respect to gender, while the treatment and control groups were balanced on demography as well as on their attitudes toward AI. Before deliberating on the proposals, participants were presented with background materials as well as a list of costs and benefits to consider.

In terms of the survey data, large majorities of participants had previously used AI. There was a statistically significant increase in these proportions after the forum. For example, in Turkey, usage rates increased from nearly 70% to 84%. In several countries, there were large increases in participants’ sense of AI’s positive benefits after deliberating, as well as a statistically significant increase in their interest. The deliberations changed participants’ opinions about a host of claims; for example, “people will feel less lonely with AI” and “more proactive [agents] are intrusive” lost approval whereas “AI agents’ capability to increase efficiency…is saving many companies a lot of time and resources” and “AI agents are helping people become more creative” gained approval. After deliberating, participants demonstrated an improved understanding of some factual aspects of AI, although the more technical aspects of this remain challenging. One example here is AI hallucinations, or rather, the generation of false or nonsensical outputs, usually because of flawed training data.
 

Image

Proposals

Participants deliberated on nineteen policy proposals. To summarize these briefly: In terms of whether and how AI remembers users’ past behaviors and preferences, participants preferred proposals that allowed users to make active choices, as opposed to this being a default setting or only being asked once. They also preferred being reminded about the ability of AI agents to personalize their experience, as well as agents being transparent with users about the tasks they complete. Participants preferred that users be educated on AI before using it, as well as being informed when AI is picking up on certain emotional cues and responding in “human-like” ways. They also preferred proposals whereby AI would ask clarifying questions before generating output. Finally, when it comes to agents helping users with real-life relationships, this was seen as more permissible when the other person was informed. Across the proposals, gender was neither a significant nor consistent determinant of how they were rated. Ultimately, the Meta Community Forum offers a model for how informed, public communication can shape AI and the ethical challenges it raises.