Paper by Michael Henry Tessler et al: “We asked whether an AI system based on large language models (LLMs) could successfully capture the underlying shared perspectives of a group of human discussants by writing a “group statement” that the discussants would collectively endorse. Inspired by Jürgen Habermas’s theory of communicative action, we designed the “Habermas Machine” to iteratively generate group statements that were based on the personal opinions and critiques from individual users, with the goal of maximizing group approval ratings. Through successive rounds of human data collection, we used supervised fine-tuning and reward modeling to progressively enhance the Habermas Machine’s ability to capture shared perspectives. To evaluate the efficacy of AI-mediated deliberation, we conducted a series of experiments with over 5000 participants from the United Kingdom. These experiments investigated the impact of AI mediation on finding common ground, how the views of discussants changed across the process, the balance between minority and majority perspectives in group statements, and potential biases present in those statements. Lastly, we used the Habermas Machine for a virtual citizens’ assembly, assessing its ability to support deliberation on controversial issues within a demographically representative sample of UK residents…(More)”.
Exploring New Frontiers of Citizen Participation in the Policy Cycle
OECD Discussion Paper: “… starts from the premise that democracies are endowed with valuable assets and that putting citizens at the heart of policy making offers an opportunity to strengthen democratic resilience. It draws on data, evidence and insights generated through a wide range of work underway at the OECD to identify systemic challenges and propose lines of action for the future. It calls for greater attention to, and investments in, citizen participation in policy making as one of the core functions of the state and the ‘life force’ of democratic governance. In keeping with the OECD’s strong commitment to providing a platform for diverse perspectives on challenging policy issues, it also offers a collection of thoughtprovoking opinion pieces by leading practitioners whose position as elected officials, academics and civil society leaders provides them with a unique vantage point from which to scan the horizon. As a contribution to an evolving field, this Discussion Paper offers neither a prescriptive framework nor a roadmap for governments but represents a step towards reaching a shared understanding of the very real challenges that lie ahead. It is also a timely invitation to all interested actors to join forces and take concerted action to embed meaningful citizen participation in policy making…(More)”.
AI in the Public Service: Here for Good
Special Issue of Ethos: “…For the public good, we want AI to help unlock and drive transformative impact, in areas where there is significant potential for breakthroughs, such as cancer research, material sciences or climate change. But we also want to raise the level of generalised adoption. For the user base in the public sector, we want to learn how best to use this new tool in ways that can allow us to not only do things better, but do better things.
This is not to suggest that AI is always the best solution: it is one of many tools in the digital toolkit. Sometimes, simpler computational methods will suffice. That said, AI represents new, untapped potential for the Public Service to enhance our daily work and deliver better outcomes that ultimately benefit Singapore and Singaporeans….
To promote general adoption, we made available AI tools, such as Pair, 1 SmartCompose, 2 and AIBots. 3 They are useful to a wide range of public officers for many general tasks. Other common tools of this nature may include chatbots to support customer-facing and service delivery needs, translation, summarisation, and so on. Much of what public officers do involves words and language, which is an area that LLM-based AI technology can now help with.
Beyond improving the productivity of the Public Service, the real value lies in AI’s broader ability to transform our business and operating models to deliver greater impact. In driving adoption, we want to encourage public officers to experiment with different approaches to figure out where we can create new value by doing things differently, rather than just settle for incremental value from doing things the same old ways using new tools.
For example, we have seen how AI and automation have transformed language translation, software engineering, identity verification and border clearance. This is just the beginning and much more is possible in many other domains…(More)”.
The Number
Article by John Lanchester: “…The other pieces published in this series have human protagonists. This one doesn’t: The main character of this piece is not a person but a number. Like all the facts and numbers cited above, it comes from the federal government. It’s a very important number, which has for a century described economic reality, shaped political debate and determined the fate of presidents: the consumer price index.
The CPI is crucial for multiple reasons, and one of them is not because of what it is but what it represents. The gathering of data exemplifies our ambition for a stable, coherent society. The United States is an Enlightenment project based on the supremacy of reason; on the idea that things can be empirically tested; that there are self-evident truths; that liberty, progress and constitutional government walk arm in arm and together form the recipe for the ideal state. Statistics — numbers created by the state to help it understand itself and ultimately to govern itself — are not some side effect of that project but a central part of what government is and does…(More)”.
It is about time! Exploring the clashing timeframes of politics and public policy experiments
Paper by Ringa Raudla, Külli Sarapuu, Johanna Vallistu, and Nastassia Harbuzova: “Although existing studies on experimental policymaking have acknowledged the importance of the political setting in which policy experiments take place, we lack systematic knowledge on how various political dimensions affect experimental policymaking. In this article, we address a specific gap in the existing understanding of the politics of experimentation: how political timeframes influence experimental policymaking. Drawing on theoretical discussions on experimental policymaking, public policy, electoral politics, and mediatization of politics, we outline expectations about how electoral and problem cycles may influence the timing, design, and learning from policy experiments. We argue electoral timeframes are likely to discourage politicians from undertaking large-scale policy experiments and if politicians decide to launch experiments, they prefer shorter designs. The electoral cycle may lead politicians to draw too hasty conclusions or ignore the experiment’s results altogether. We expect problem cycles to shorten politicians’ time horizons further as there is pressure to solve problems quickly. We probe the plausibility of our theoretical expectations using interview data from two different country contexts: Estonia and Finland…(More)“.
We’ve Got a Big Problem
Blog by Daro: “There is a problem related to how we effectively help people receiving social services and public benefit programs. It’s a problem that we have been thinking, talking, and writing about for years. It’s a problem that once you see it, you can’t unsee it. It’s also a problem that you’re likely familiar with, whether you have direct experience with the dynamics themselves, or you’ve been frustrated by how these dynamics impact your work. In February, we organized a convening at Georgetown University in collaboration with Georgetown’s Massive Data Institute to discuss how so many of us can be frustrated by the same problem but haven’t been able to really make any headway toward a solution.
For as long as social services have existed, people have been trying to understand how to manage and evaluate those services. How do we determine what to scale and what to change? How do we replicate successes and how do we minimize unsuccessful interventions? To answer these questions we have tried to create, use, and share evidence about these programs to inform our decision-making. However – and this is a big however – despite our collective efforts, we have difficulty determining whether there’s been an increase in using evidence, or most importantly, whether there’s actually been an improvement in the quality and impact of social services and public benefit programs…(More)”.
Ensuring citizens’ assemblies land
Article by Graham Smith: “…the evidence shows that while the recommendations of assemblies are well considered and could help shape more robust policy, too often they fail to land. Why is this?
The simple answer is that so much time, resources and energy is spent on organising the assembly itself – ensuring the best possible experience for citizens – that the relationship with the local authority and its decision-making processes is neglected.
First, the question asked of the assembly does not always relate to a specific set of decisions about to be made by an authority. Is the relevant policy process open and ready for input? On a number of occasions assemblies have taken place just after a new policy or strategy has been agreed. Disastrous timing.
This does not mean assemblies should only be run when they are tied to a particular decision-making process. Sometimes it is important to open up a policy area with a broad question. And sometimes it makes sense to empower citizens to set the agenda and focus on the issues they find most compelling
The second element is the failure of authorities to prepare to receive recommendations from citizens.
One story is where the first a public official knew about an assembly was when its recommendations landed on their desk. They were not received in the best spirit.
Too often assemblies are commissioned by enthusiastic politicians and public officials who have not done the necessary work to ensure their colleagues are willing to give a considered response to the citizens’ recommendations. Too often an assembly will be organised by a department or ministry where the results require others in the authority to respond – but those other politicians and officials feel no connection to the process.
And too often, an assembly ends, and it is not clear who within the public authority has the responsibility to take the recommendations forward to ensure they are given a fair hearing across the authority.
For citizens’ assemblies to be effective requires political and administrative work well beyond just organising the assembly. If this is not done, it is not only a waste of resources, but it can do serious damage to democracy and trust as those citizens who have invested their time and energy into the process become disillusioned.
Those authorities where citizens’ assemblies have had meaningful impacts are those that have not only invested in the assembly, but also into preparing the authority to receive the recommendations. Often this has meant continuing support and resourcing for assembly members after the process. They are the best advocates for their work…(More)”
Can LLMs advance democratic values?
Paper by Seth Lazar and Lorenzo Manuali: “LLMs are among the most advanced tools ever devised for analysing and generating linguistic content. Democratic deliberation and decision-making involve, at several distinct stages, the production and analysis of language. So it is natural to ask whether our best tools for manipulating language might prove instrumental to one of our most important linguistic tasks. Researchers and practitioners have recently asked whether LLMs can support democratic deliberation by leveraging abilities to summarise content, as well as to aggregate opinion over summarised content, and indeed to represent voters by predicting their preferences over unseen choices. In this paper, we assess whether using LLMs to perform these and related functions really advances the democratic values that inspire these experiments. We suggest that the record is decidedly mixed. In the presence of background inequality of power and resources, as well as deep moral and political disagreement, we should be careful not to use LLMs in ways that automate non-instrumentally valuable components of the democratic process, or else threaten to supplant fair and transparent decision-making procedures that are necessary to reconcile competing interests and values. However, while we argue that LLMs should be kept well clear of formal democratic decision-making processes, we think that they can be put to good use in strengthening the informal public sphere: the arena that mediates between democratic governments and the polities that they serve, in which political communities seek information, form civic publics, and hold their leaders to account…(More)”.
Data-driven decisions: the case for randomised policy trials
Speech by Andrew Leigh: “…In 1747, 31-year-old Scottish naval surgeon James Lind set about determining the most effective treatment for scurvy, a disease that was killing thousands of sailors around the world. Selecting 12 sailors suffering from scurvy, Lind divided them into six pairs. Each pair received a different treatment: cider; sulfuric acid; vinegar; seawater; a concoction of nutmeg, garlic and mustard; and two oranges and a lemon. In less than a week, the pair who had received oranges and lemons were back on active duty, while the others languished. Given that sulphuric acid was the British Navy’s main treatment for scurvy, this was a crucial finding.
The trial provided robust evidence for the powers of citrus because it created a credible counterfactual. The sailors didn’t choose their treatments, nor were they assigned based on the severity of their ailment. Instead, they were randomly allocated, making it likely that difference in their recovery were due to the treatment rather than other characteristics.
Lind’s randomised trial, one of the first in history, has attained legendary status. Yet because 1747 was so long ago, it is easy to imagine that the methods he used are no longer applicable. After all, Lind’s research was conducted at a time before electricity, cars and trains, an era when slavery was rampant and education was reserved for the elite. Surely, some argue, ideas from such an age have been superseded today.
In place of randomised trials, some put their faith in ‘big data’. Between large-scale surveys and extensive administrative datasets, the world is awash in data as never before. Each day, hundreds of exabytes of data are produced. Big data has improved the accuracy of weather forecasts, permitted researchers to study social interactions across racial and ethnic lines, enabled the analysis of income mobility at a fine geographic scale and much more…(More)”
‘Evidence banks’ can drive better decisions in public life
Article by Anjana Ahuja: “Modern life is full of urgent questions to which governments should be seeking answers. Does working from home — WFH — damage productivity? Do LTNs (low-traffic neighbourhoods) cut air pollution? Are policy ideas that can be summed up in three-letter acronyms more palatable to the public than those, say Ulez, requiring four?
That last one is tongue-in-cheek — but the point stands. While clinical trials can tell us reasonably confidently whether a drug or treatment works, a similar culture of evaluation is generally lacking for other types of intervention, such as crime prevention. Now research funders are stepping into the gap to build “evidence banks” or evidence syntheses: globally accessible one-stop shops for assessing the weight of evidence on a particular topic.
Last month, the Economic and Social Research Council, together with the Wellcome Trust, pledged a total of around £54mn to develop a database and tools that can collate and make sense of evidence in complex areas like climate change and healthy ageing. The announcement, Nature reports, was timed to coincide with the UN Summit of the Future, a conference in New York geared to improving the world for future generations.
Go-to repositories of good quality information that can feed the policy machine are essential. They normalise the role of robust evidence in public life. This matters: the policy pipeline has too often lacked due diligence. That can mean public money being squandered on ineffective wheezes, or worse.
Take Scared Straight, a crime prevention scheme originating in the US around 40 years ago and adopted in the UK. It was designed to keep teens on the straight and narrow by introducing them to prisoners. Those dalliances with delinquents were counterproductive. A review showed that children taking part were more likely to end up committing crimes than those who did not participate in the scheme…(More)”.