Much of the agenda focused on specific proposals answering ten basic questions, including:

• Which sources should AI chatbots draw information from?
• Should AI chatbots use the user’s past conversations to improve user experience?
• Should AI chatbots use the user’s online activity to personalize interactions?
• Should AI chatbots be designed to be human-like?
• Should users be allowed to use AI chatbots for romantic relationships?

The deliberators in each country were convened on the Stanford Online Deliberation Platform, developed by the Stanford Crowdsourced Democracy Team. This platform is AI-assisted and moderates video-based discussions in small groups of ten. 

Participants noted a generally positive impact of AI, with an increase in positive perceptions post-forum. "This shift in viewpoint, especially notable among those previously unfamiliar with AI, underscores the value of informed discussion in shaping public opinion," remarked Alice Siu, Senior Research Scholar at FSI and Associate Director of the Stanford Deliberative Democracy Lab.

The forum's methodical approach, with deliberations in 166 small groups and a separate control group for each country, ensured that the outcomes accurately reflected the effects of the deliberative process. Findings indicated that users and non-users of AI began to bridge their differences, aligning closer in their attitudes towards AI after the forum.

Siu also emphasized the significance of the event’s feedback, noting that "the high ratings from participants affirm the forum’s success in making complex discussions accessible and relevant. It is an encouraging sign for the future of democratic engagement with technology."

The Stanford Deliberative Democracy Lab is dedicated to deepening our understanding of deliberative democracy processes and their application to critical contemporary issues like AI. The insights from this forum will inform ongoing research and policy recommendations, ensuring that AI evolves in a manner that considers diverse values and voices across the globe.

The project report is available below:

For detailed information on the forum’s conclusions or forthcoming projects, please reach out to James Fishkin (Director, Stanford Deliberative Democracy Lab) or Alice Siu (Associate Director, Stanford Deliberative Democracy Lab) or visit the Deliberative Democracy Lab website.

There are a number of ways to run a legitimate election. But the U.S. has learned in recent years, and Brazil learned in recent weeks, that it’s not always simple.

There are technical mechanics and processes of how votes are cast, collected and counted. But those are ultimately less important than the agreement – among opposing parties, and across a society – to abide by the results of those processes.

In 2020, President Donald Trump alleged, without evidence, that election fraud in several states had caused him to lose. A number of audits in various states found no evidence that irregularities in voting or vote counting processes had any effect on the outcome of balloting in those states.

Some of these results were later challenged in lawsuits seeking to alter the results of the election, and in every case, the election’s outcome was determined to be accurate.

Continue reading at theconversation.com

Since the pandemic started, one in 10 people living in Brazil has contracted COVID-19, and more than 600,000 deaths have been reported. Despite these staggering numbers, Jair Bolsonaro, the president of Brazil, has consistently downplayed the threat of the pandemic. The Brazilian Senate has had enough.

On October 26, seven out of the 11 members in a Senate committee voted in favor of recommending nine charges against the Brazilian president. Over the last six months, a Brazilian Senate Parliamentary Inquiry Committee composed of 11 Senators investigated the federal government’s management of the pandemic. 

Read the rest at nacla

There’s a great deal of support for community policing, but not a lot of concrete data on whether it works. What makes this a challenging issue to study?

Christia: Building trust between police and the citizens they are tasked to protect is at the core of community policing. As such, an important challenge lies with identifying the right measures to capture this often-complex police-citizen interaction. This was even more of a pronounced challenge in our study as we conducted six coordinated experiments across a diverse set of sites in the Global South in Brazil, Colombia, Liberia, Pakistan, the Philippines and Uganda.

To make progress in understanding the impacts of community policing, we needed to develop a set of common strategies for the police to implement that made sense in each national context, which we call locally appropriate community policing interventions. We also needed to agree upon a shared research design across countries and to introduce common outcome measures to ensure that we were looking at the impacts of these programs in similar ways. This approach to launching coordinated, multi-site, randomized controlled trials across contexts has been pioneered by the organization that sponsored this work, Evidence in Governance and Politics (EGAP).

Your team partnered with six communities across the Global South in Brazil, Colombia, Liberia, Pakistan, the Philippines and Uganda. Based on your research, what evidence did you find for or against the use of community policing practices?

Blair: We find that community policing doesn’t live up to its promise when implemented in the Global South. Community policing doesn’t build trust between citizens and police, it doesn’t lead to citizens to share the kinds of tips and information with police that might improve police efficiency, and, perhaps not surprisingly then, it does not lead to lower crime. This disappointing result was apparent across all six contexts and for all of the primary outcomes we measured.

Is there an alternative to community policing, or ways to reform these systems, that would make them more efficacious at creating the desired outcomes?

Weinstein: We carefully examined each of the six contexts, including through interviews with the police agencies and the research teams, to make sense of this null result. We identified three primary constraints that may have impeded the implementation of community policing: (a) a lack of prioritization of these new practices by police leadership (b) the rotation to new posts of police officers who had championed the effort and were trained to implement it and (c) limited resources to follow up on the concerns raised by citizens.

The bottom line is that community policing isn’t positioned to deliver increased trust and collaboration in environments with limited incentives and resources to enable police to change their behavior. Our conclusion is that community policing should be seen as an incremental reform that can make a difference in well-resourced police departments with strong incentives to be responsive to citizen concerns. But when those conditions are absent, an incremental approach can’t deliver. More systemic reforms are required.

How does the data from your work fit into broader issues of equity, just representation, and racism that communities across the world continue to grapple with?

Blair: In many ways community policing appears to be the ideal policy for this moment, where so many are demanding that police abuse be reduced while also reducing crime victimization. Community policing is meant to do both, constructing a virtuous cycle between citizen-police cooperation, trust, and crime reduction. Our null results sound a note of caution: it may not be so simple. We observed big barriers to implementing this shift in policing, and barriers that likely affect other incremental policies. To address equity in the way governments enforce the law, we may need more systematic changes to how we organize the police and hold them accountable.