What is the best democracy ROI
This map is a direct call to funders looking to invest at the frontiers of AI governance and democratic decision-making.
Strategic funding requires knowing where investment creates the most leverage. The map identifies the capabilities limiting deliberation’s scale and impact, the capabilities that, once made possible, built, and adopted, will unlock many new paths for how we collectively make decisions around AI.
This page includes sections for: Fundamental research gaps, low-hanging fruit and high impact niches.
Fundamental research gaps
Clearing a few key bottlenecks has the potential to unlock significant uplift through flywheels and spillover improvements in many capabilities.
Table
Tags
Urgent
Deliberative processes that can be tested and refined before implementation with real participants
Urgent
Long-term effects of participation that can be tracked on individuals and their networks
Urgent
Post-hoc verification that can establish that outcomes were not unduly influenced.
Urgent
Assembly designs that are robust to both internal and external manipulation attempts.
Urgent
Product "Neutrality dashboards" that provide real-time feedback to facilitators and organizers.
Urgent
Governments that can legally bind themselves to the decisions of a deliberative process
Urgent
Private organizations that can legally bind themselves to the decisions of a deliberative process
Urgent
Real-time monitoring and correction of neutrality violations that is feasible.
Urgent
Participants who learn as much as possible in the time available.
Urgent
Ensure that commitments made by commissioning authorities to act on process outputs stick
Urgent
Ensure that commitments are sufficiently adaptable to changing circumstances
Urgent
Recruitment strategies that accommodate poor databases for identifying and accessing people locations with no mail access or poor access to the internet.
Urgent
Digital tools that enhance facilitation quality and capacity.
Urgent
That neutrality can be defined and measured across different assembly components.
Urgent
Preserve human elements that make deliberation legitimate and valuable
Urgent
Participants who are able to accurately understand the potential impacts of their decisions.
Urgent
Metrics of quality of deliberation that can be measured in real time, enabling facilitators to make adaptive process interventions
Urgent
Deliberations that can be captured faithfully and unobtrusively with full participant consent and ethical protection
Urgent
Preference transformation and participant learning that can be tracked throughout the process
Urgent
Scenarios that can be reliably enumerated to inform deliberations.
Urgent
Process outcomes that can be empirically measured and compared across contexts, processes and systems, enabling evidence-based improvements
Urgent
Legal frameworks that provide definitions of different degrees of bindingness with timelines and enforcement triggers
Urgent
Safe, technically binding decision-making systems
Urgent
Tools that track the quality of individual and group learning within the deliberative process
Urgent
Sequences of micro-processes that can be easily connected to create larger processes
Urgent
Designers who understand the validity boundaries and appropriate use cases for deliberative process simulation
Urgent
Response rates that enable selection algorithms to accurately select panels that represent the whole population by mitigating the impacts of self-selection biases.
Urgent
Internal barriers that are removed within organizations to be able to make and act on commitments
Urgent
Enumerated scenarios that can be effectively and fairly integrated into deliberations.
Urgent
There are cheap and efficient ways of recruiting participants that can account for differences in response rates to different types of invitation methods (mail phone door-knocking etc).
Urgent
Manipulation attempts that can be reliably detected and prevented across different stages of the assembly process.
Urgent
Low-hanging fruit
Democratic innovations are significantly underfunded, resulting in large amounts of low-hanging fruit.
Activate learning
Urgent Informedness
Ability for diverse participants to efficiently and effectively learn relevant information, such that they can actively apply their learnings in the process.
Bind technically
Urgent Bindingness
Ability to ensure the relevant technical systems act in directed ways (e.g., via model alignment, control, cryptographic enforcement of outputs, physical limitations, etc.). Caveat: Irrevocable direct binding to outputs should generally be implemented with checks and balances.
Build process workflows
Urgent Adaptability
Ability to construct process workflows that achieve intended outcomes in given contexts.
Curate context
Informedness
Ability to provide complete context to participants, including things like background information, subject matter fundamentals, relevant considerations, tradeoffs, and possible options.
Enumerate scenarios
Urgent Informedness
Ability to generate lists of likely scenarios, including edge cases, in which decisions will be applied, to help participants better understand the issue space.
Evaluate processes
Urgent Measurability
Ability to measure desired outcomes to compare methods, processes and systems.
Forecast impacts
Urgent Informedness
Ability to effectively and easily model complex systems, to help participants understand the potential impacts of different decisions.
Gather process data
Urgent Measurability
Ability to gather process data in a cheap, reliable, accessible manner.
Handle challenges
Robustness
Ability to withstand changing contexts and less-than-ideal conditions.
Include voiceless perspectives
Representativeness
Ability to fairly include the perspectives of those that are not represented in the process, including people who are not present (future generations, young people or other representation constraints), and non-human entities (natural phenomena or animals).
Integrate wider-public
Informedness
Ability to provide those not in the room deliberating with opportunities to constructively and fairly contribute input into the process.
Make verifiable
Legibility
Ability for integrity of the process to be verified and audited.
Manage data
Process Speed
Ability to manage, route and surface data produced by the process throughout the process.
Manage subsidiarity
Scalability
Ability to host decentralized processes simultaneously or sequentially and productively distill them into one central process.
Navigate ambiguity
Bindingness
Ability to ensure that, given potential ambiguity of decisions, the authority takes actions as close to the intended ones as possible.
Navigate conflict
Robustness
Ability to address, resolve and navigate conflicts that emerge within the process.
Optimize run-time
Process Speed
Ability to run time-minimal processes subject to performance and reliability thresholds.
Produce adaptable outputs
Substantiveness
Ability for final outputs to be adaptable to changing contexts while retaining clear intended outcomes and specificity.
Reach participants
Urgent Representativeness
Ability to reach potential participants (e.g., to mitigate biases around self-selection, who is reachable, etc.).
Represent complexity
Substantiveness
Ability for final outputs to be nuanced, concrete, decisive, and comprehensive.
Simulate participation
Representativeness
Ability to simulate the interactions and decisions of actors (e.g., participants, stakeholders, facilitators, experts), subprocesses, or entire processes (e.g., for rapid process iteration).
Simulate prototyping
Urgent Learning Speed
Ability to run trials that are good enough to learn from, and fast enough to enable rapid testing of new methods and process comparisons.
Work transnationally
Scalability
Ability to run deliberative processes at the transnational level by navigating challenges such as legitimacy, logistics, and cultures.
Neglected work
Many actors are working on the same, obvious problems, leaving many key challenges comparatively neglected.
Table
Neglectedness
High Extreme
Activate learning
Urgent Informedness
Ability for diverse participants to efficiently and effectively learn relevant information, such that they can actively apply their learnings in the process.
Bind technically
Urgent Bindingness
Ability to ensure the relevant technical systems act in directed ways (e.g., via model alignment, control, cryptographic enforcement of outputs, physical limitations, etc.). Caveat: Irrevocable direct binding to outputs should generally be implemented with checks and balances.
Collectivize data
Learning Speed
Ability to make data open and easily available to researchers.
Curate context
Informedness
Ability to provide complete context to participants, including things like background information, subject matter fundamentals, relevant considerations, tradeoffs, and possible options.
Enumerate scenarios
Urgent Informedness
Ability to generate lists of likely scenarios, including edge cases, in which decisions will be applied, to help participants better understand the issue space.
Evaluate processes
Urgent Measurability
Ability to measure desired outcomes to compare methods, processes and systems.
Forecast impacts
Urgent Informedness
Ability to effectively and easily model complex systems, to help participants understand the potential impacts of different decisions.
Gather process data
Urgent Measurability
Ability to gather process data in a cheap, reliable, accessible manner.
Handle challenges
Robustness
Ability to withstand changing contexts and less-than-ideal conditions.
Integrate transnationally
Integration
Ability to integrate with transnational and interorganizational systems.
Make verifiable
Legibility
Ability for integrity of the process to be verified and audited.
Manage data
Process Speed
Ability to manage, route and surface data produced by the process throughout the process.
Maximize neutrality
Urgent Robustness
Ability to increase, demonstrate, or measure the neutrality of key aspects of a process.
Navigate ambiguity
Bindingness
Ability to ensure that, given potential ambiguity of decisions, the authority takes actions as close to the intended ones as possible.
Navigate conflict
Robustness
Ability to address, resolve and navigate conflicts that emerge within the process.
Produce adaptable outputs
Substantiveness
Ability for final outputs to be adaptable to changing contexts while retaining clear intended outcomes and specificity.
Produce implementable outputs
Substantiveness
Ability to produce outputs in immediately actionable forms (e.g. policies, budgets, AI constitutions, town plans etc.)
Reach participants
Urgent Representativeness
Ability to reach potential participants (e.g., to mitigate biases around self-selection, who is reachable, etc.).
Represent complexity
Substantiveness
Ability for final outputs to be nuanced, concrete, decisive, and comprehensive.
Resist manipulation
Urgent Robustness
Ability to resist manipulation that would decrease trustworthiness, legitimacy or unfairly influence the outcome.
Simulate participation
Representativeness
Ability to simulate the interactions and decisions of actors (e.g., participants, stakeholders, facilitators, experts), subprocesses, or entire processes (e.g., for rapid process iteration).
Simulate prototyping
Urgent Learning Speed
Ability to run trials that are good enough to learn from, and fast enough to enable rapid testing of new methods and process comparisons.