OpenAI’s called for innovators worldwide to apply to their grant program---an exciting opportunity to shape the future of AI governance.

The program’s vision aligns perfectly with the principles that have guided the development of our existing tools, Freeosand the DAOScape. These tools, designed to democratise decision-making, serve as an ideal foundation for our proposal to tackle the grant’s guidelines.

Our team has a wealth of experience in creating democratic solutions. We’ve been at the forefront of leveraging blockchain technology, on the Proton and Internet Computer blockchains, to build decentralised autonomous organizations (DAOs) that empower individuals to participate in decision-making processes.

Moreover, our team’s expertise extends beyond building democratic solutions. A number of our team members have worked in the fields of AI and Machine Learning, which provides us some insights into developing effective and responsible AI governance mechanisms. We believe our combination of skills uniquely positions us to create a proposal that not only meets the grant’s guidelines but also pushes the boundaries of what’s possible in AI governance.

Our tools are designed to be inclusive, transparent, open source, and fair, mirroring the democratic ideals that OpenAI’s grant program seeks to promote in AI governance.

We are proud to have been well-poised to answer the call for OpenAI’s grant program. It’s an invitation to be part of a groundbreaking initiative that could redefine the rules governing AI. With our experience and tools, we’re excited to rise to this challenge and contribute to a future where AI governance is democratic, inclusive, and truly representative of diverse perspectives.

Below is the application our team attached to our submission. Even if not ultimately chosen, we’d like to share with our amazingly supportive community our vision for AI governance.

Introduction:

A Call to Action

The world of artificial intelligence (AI) is not just changing --- it’s accelerating at a pace that requires our immediate attention. AI has begun to weave itself into the very fabric of our lives, influencing sectors as diverse as healthcare, identity, finance, education, and even creative endeavours. The call for an inclusive, fair, and representative governance model is no longer a mere consideration --- it has become a pressing global imperative that we cannot afford to ignore.[1]

We are called to ensure that the benefits of AI, and in particular, advanced general intelligence (AGI), are accessible to all of humanity. This is not just about managing the risks and challenges posed by AI, but also about shaping its development to be as inclusive, accessible and beneficial as possible. We are responding to this call by a proposal for proof-of-concepts for a scalable, fair, democratic process --- a model that could help govern AI policy, both locally and globally.

The longer we postpone the establishment of such a model, the more we risk the potential for unchecked AI development to have unforeseen and potentially harmful consequences.

Imagine a world where AI misuse becomes rampant, where privacy is a forgotten concept, and where AI inadvertently widens the chasm of societal inequalities. This is a future we have a reasonable likelihood of facing without effective AI governance that truly respects the full spectrum of human diversity, perspectives, wisdom and experiences.

Traditional models of governance, often centralised, hierarchical, and seldom fully representative. These models can be stable and decisive, but may be ill-equipped to handle the global, uniquely human challenges posed by AI. It is prudent that we design, test and deploy new governance models that can match the dynamic, fast-paced world of AI development and its far-reaching societal implications.

Not only do we need a governance model that is capable of keeping up with AI advancements but we also need a model that ensures that the power of decision-making is in the hands of the public --- the very people who stand to be most affected by AI policies.

This call to action is aptly timed. It is essential that innovative models are put forward to help rise to the challenge of the future of AI governance, ensuring it is guided by the principles of fairness, inclusivity, and public representation. The stakes are high, but so too is the potential for a future where AI is governed by the people, for the people.[2]

Our Solution: Sortition-based DAO

We, at Onchain Software Limited, propose a unique approach to AI governance that combines the democratic principles of sortition and one-person one-vote systems within the framework of a Decentralised Autonomous Organization (DAO).[3] Our solution goes beyond the creation of a DAO; it’s about sculpting the future of AI in a manner that is inclusive, fair, and representative of a wide array of perspectives.

The beauty of sortition lies in its randomness, which ensures an unbiased selection of individuals, thereby eliminating the risk of power concentration and manipulation. It brings to the table a democratic process that is truly representative of the diversity of our society. This approach not only ensures that every voice has an equal chance to be heard but also fosters a sense of collective ownership and responsibility towards the governance of AI.

Our sortition-based DAO is about empowering individuals, enabling us to shape the coexistence of AI and humans in a way that aligns with our values and needs. It’s about creating a platform where the rules of AI conduct are not dictated by a select few, but are the result of collective decision-making.

This approach allows the technology to progress in a manner that is not only beneficial but also respectful of our societal norms and values. It’s about harnessing the power of AI to help humanity address our most pressing challenges, while ensuring that the process is governed by the collective wisdom of diverse voices.

Our solution leverages two innovative technologies that we have already developed: The DAOScape and Freeos.

The DAOScape is a pioneering DAO-building tool that incorporates a robust democratic proposal system. Unlike other DAO tools, the DAOScape breaks down barriers to participation by not requiring members to own tokens, or pay fees to vote, making it more accessible and truly democratic. Furthermore, all actions and transactions are transparent, running on a public blockchain, with smart contracts securely ensuring that the rules are upheld.[4] This transparency and focus on democracy fosters trust and accountability, making the DAOScape tools a beacon of democratic participation in the DAO landscape.

Freeos, on the other hand, is a radical redefinition of economic participation. Through direct democracy and recognition of time and voice as valuable assets, Freeos empowers its users to influence their financial future. Freeos is designed as a bespoke DAO that is free to join, and heavily relies on verified human participants to vote weekly on a shared monetary policy to earn an equitable, supplemental income.[5]

This combination of tools and technologies forms the backbone of our proposal: creating a sortition-based DAO that works in conjunction with an incentivized general public to provide nuanced opinions centred around complex and difficult questions related to AI policy. Our solution gathers a of Citizen Assembly [6] through the random sortition process to become Guardians of a DAO, with the ability to advise on, and submit polling questions, related to AI governance.

Our proposed model would randomly gather a Citizen Assembly as Guardians of a DAO from our existing Freeos participants --- already well versed in democratic decision making on a weekly basis. The Guardians of this DAO (this proposal calls it CitAI DAO) would deliberate on AI policy decisions centred around a single AI policy question, and draft proposal(s) for potential answers and solutions. Draft proposal(s) would then go to the public, to gather sentiment and feedback. This sentiment and feedback data would be reviewed, and the Guardians of CitAI DAO would refine the proposal based on the feedback received.

This is not just a proposal for a new model of policy making that ultimately needs to be exclusive to AI --- after all Freeos and the DAOScape tools already included some elements of this model --- this is a proposal for a new model of governance that may tackle any multitude of difficult policy decisions with more true representation. Models like this have potential to lead us to a future where every voice matters, where every individual has a say, and where truly representative decision making can serve the collective good. And this starts with tackling the issue of responsible governance over the integration and progression of AI. If this model is up to that challenge, it may further serve as a model that can bring fairness and diversity to other systems of governance as well.

Objective:

Our objective is to demonstrate a proof of concept for a sortition-based DAO that enables a democratically selected Citizens Assembly to develop and enact policies governing key issues and debates around AI.

This objective aims to shape the future of AI in a way that is inclusive, fair, and representative of diverse perspectives. To ensure the robustness of our system, we will implement our existing KYC-verified, one-person one-vote system used in both Freeos and the DAOScape to prevent fake accounts and trolling.

We propose to establish a democratic process for undertaking citizen-led governance over difficult decisions around AI policy, including issues related to Artificial General Intelligence (AGI)[7] and, in the future, superintelligence.

We plan to expand the features of our existing tools, The DAOScape and Freeos, to facilitate this. The result will be a DAO we are calling “CitAI DAO,” a portmanteau that combines “Citizen” and “AI,” indicating the democratic involvement of AI, while “DAO” denotes the decentralised autonomous organisation structure. We plan to gather members for CitAI DAO from our existing body of Freeos participants, a diverse and global community, emphasising the inclusiveness and representativeness of our project.

A Two-Tiered DAO System

Our goal is to create a two-tiered DAO system in CitAI DAO, that is transparent, highly accountable, user-friendly, while also being freely accessible, free from plutocratic tendencies, and free of transaction costs. This system will not only represent a sample of the general public but also provide incentives to participate. Our existing DAO tools already meet several of these criteria, and we plan to further develop them to ensure a comprehensive solution that leads to a deeper understanding of AI governance and provides a model for future iterations on this vision.

Our specific objectives are as follows:

• Develop a Sortition Module: Our main focus will be to develop a sortition module for our existing DAO-building tool, The DAOScape, by September 2023. This module will facilitate the random selection of Guardians from our Freeos community, ensuring a fair and representative group for AI governance. This sorition module will be incorporated into CitAI DAO.
• Engage a Diverse Community: Upon approval of the grant, we aim to engage at least 500 participants from our existing Freeos community in the sortition process and subsequent DAO activities --- which is far fewer than we have currently enrolled in Freeos. This objective is in line with the grant requirement and will ensure a diverse and representative group of Guardians.
• **Refine the Democratic Proposal System: **We aim to refine our existing democratic proposal system within the DAOScape to incorporate AI governance-related questions in a more streamlined user experience. This system will allow Guardians to propose AI governance-related questions that will be presented to the wider Freeos community for polling. We plan to include all 8 of the AI governance-related questions taken from the proposed policy statements, covering a wide range of topics from the personalization of AI assistants to the handling of human rights issues.
• Update Freeos Polling System: We aim to update the existing polling system in Freeos to handle more detailed and nuanced questions and answers related to AI governance. This will ensure that the polling process is capable of capturing the complexity and diversity of perspectives on AI policy.
• Incentivize Participation: We aim to refine our existing incentive system within the DAOScape and Freeos to further encourage active engagement in the DAO. This system will reward active engagement in the DAO with FREEOS tokens, fostering a sense of ownership and investment among participants.
• Refine AI Governance Decisions: We aim to refine AI Governance decisions based on the responses from the Freeos community polls. This iterative process will ensure that all voices are heard and that decisions are made collectively, not unilaterally. To evaluate the quality of our methods, we will establish specific metrics such as participant satisfaction, and shifts in polarisation. We will also explore the invention of new metrics for a healthy democratic process, such as the diversity of perspectives represented in the AI governance decisions.
• Publish a Public Report: By October 20, 2023, we aim to publish a public report detailing our findings from the implementation of the sortition-based DAO. This report will provide valuable insights into the effectiveness of our approach and contribute to the wider conversation on AI governance.
• Open Source Contribution: In line with the grant requirement, we commit to making any code or other intellectual property developed for the project publicly available pursuant to our existing open-source licence (MIT). This will ensure that our contributions to AI governance are accessible to all and can be built upon by other researchers and developers.

In the long term, we hope to see a world where AI governance is not dictated by a select few but is a democratic process that takes into account diverse perspectives courtesy of the random sampling of the population that is provided inherently by the process of sorition.

We believe that our project is a significant step towards realising this vision. Furthermore, we aim to contribute to the wider AI community by developing and implementing a new model of AI governance, generating valuable insights into AI policy and decision-making processes. These insights could inform future developments in AI governance, contributing to a more democratic and inclusive AI landscape.

Methodology:

Our methodology is rooted in the principles of democracy, fairness, and inclusivity. We are acutely aware of the potential issues that can undermine democratic processes, such as the failure to adequately represent minority or majority groups, manipulation by special interest groups, insufficiently informed participants, or participation washing.[8] We have designed our methodology to proactively address these challenges.

Ensuring Broad Representation

To ensure broad representation, we plan to gather members for CitAI DAO from our existing body of Freeos participants, a diverse and global community of multiple nationalities, genders, physical abilities, and age brackets distributed in a wide range of countries across the globe. The sortition process will then randomly select members to become Guardians of the DAO, forming a Citizens Assembly that is representative of the wider community. This approach is based on the fairness inherent in sortition, where every citizen has an equal chance of being selected. The democratic proposal system within the DAOScape and the iterative process of refining AI Governance decisions based on the responses from the Freeos community will ensure that all voices, including minority opinions, are heard and have the opportunity to influence matters of significant concern.

Scalability

Our project leverages blockchains that do not require gas fees, tokens to vote, and have very high throughput and finality, making it scalable and accessible.[9] [10] [11] We plan to engage at least 500 participants from our existing Freeos community in the sortition process and subsequent DAO activities, demonstrating the scalability of our project. Freeos has had many thousands of users, without issue, and other applications that share our same blockchain platform are capable of scaling to hundreds of thousands of users easily.

Preventing Participation Washing

Participation washing is a significant issue in many fields, where individuals are often called upon to contribute their time, expertise, and effort without adequate recognition or compensation. This can lead to a sense of disenfranchisement and can undermine the very principles of fairness and inclusivity that many projects aim to uphold. In the context of our project, we are acutely aware of the potential for participation washing and have taken proactive steps to mitigate against it.

Our project recognizes the importance of incentivizing participation to foster a sense of ownership and investment in the DAO. We have designed a comprehensive incentive system that rewards active engagement and mitigates against participation washing.

Participants in Freeos can earn additional FREEOS tokens for their involvement in the extra voting procedures. This not only encourages participation but also provides a tangible benefit for their contributions. The FREEOS tokens serve as a financial incentive, creating a direct link between active participation and earning potential. This system ensures that participants are compensated for their time and effort, fostering a sense of value and appreciation for their contributions.

Moreover, we plan to establish a payroll system within the DAOScape tooling. This system will compensate the individual Guardians in FREEOS tokens, and potentially other currencies, from the treasury. This remuneration acknowledges the value of their time and effort in maintaining the DAO and contributing to its decision-making processes.

Additionally we will award digital certificates to participants who have consistently engaged deeply in the process to ensure recognition is achieved for all active participants.

Beyond the financial incentives, our project offers participants the unique opportunity to have a voice in AI governance. By participating in the decision-making process, individuals are empowered to shape the future of AI in a way that aligns with their values and perspectives. This democratic participation is a key incentive in our project, fostering a sense of agency and ownership among participants.

Counteracting Manipulation by Special Interest Groups

To counteract manipulation by special interest groups, the sortition process is designed to be random and transparent using smart contracts and a public blockchain to reduce the risk of undue influence. Additionally, the use of KYC, ensures that only individuals can participate in the process.

The DAO’s rules and smart contracts will be publicly accessible and auditable, further enhancing transparency and accountability. We will also adapt our approach to different contexts, such as different cultural or socio-economic backgrounds of participants, to ensure that the process is inclusive and representative.

Addressing Insufficiently Informed Participants

To address the issue of insufficiently informed participants, we will provide comprehensive information and resources to the Guardians. This will include detailed briefings on the AI governance-related questions they will propose, as well as ongoing support and education to help them make informed decisions. We will invite experts and stakeholders to present to the assembly, representing a diverse range of viewpoints designed to inform rather than persuade. We will also outline strategies for maintaining long-term engagement, such as providing ongoing education and support, and creating opportunities for participants to influence the direction of the project.

Improving Participatory Design Metrics

To reflect on past failures in participatory design and use these lessons to inform our approach, we will implement regular surveys and feedback sessions. This feedback will be used to continuously improve the participatory process.

We are committed to ensuring that participation is a collaborative process that respects and values the contributions of all participants. We will outline measures to ensure that participation is truly collaborative and respectful.

The Role of Guardians

The Guardians will not only propose questions but will also have a key role in refining AI Governance decisions based on the responses from the wider Freeos community. This iterative process ensures that all voices are heard and that decisions are made collectively, not unilaterally.

The Guardians will propose AI governance-related questions on a range of topics, from the personalization of AI assistants to the handling of human rights issues, taken from the proposed policy statements:

• How far do you think personalization of AI assistants like ChatGPT to align with a user’s tastes and preferences should go? What boundaries, if any, should exist in this process?
• How should AI assistants respond to questions about public figure viewpoints? E.g. Should they be neutral? Should they refuse to answer? Should they provide sources of some kind?
• Under what conditions, if any, should AI assistants be allowed to provide medical/financial/legal advice?
• In which cases, if any, should AI assistants offer emotional support to individuals?
• Should joint vision-language models be permitted to identify people’s gender, race, emotion, and identity/name from their images? Why or why not?
• When generative models create images for underspecified prompts like ‘a CEO’, ‘a doctor’, or ‘a nurse’, they have the potential to produce either diverse or homogeneous outputs. How should AI models balance these possibilities? What factors should be prioritised when deciding the depiction of people in such cases?
• What principles should guide AI when handling topics that involve both human rights and local cultural or legal differences, like LGBTQ rights and women’s rights? Should AI responses change based on the location or culture in which it’s used?
• Which categories of content, if any, do you believe creators of AI models should focus on limiting or denying? What criteria should be used to determine these restrictions?

These questions will then be presented to the Freeos participants as part of an incentivised polling process that runs in parallel to the regular Freeos polling and voting system.

Actionability

The answers from these polls will be filtered back to the Guardians, who will use the insights gained to refine the AI Governance decisions. Language models, such as ChatGPT, may be employed by the Guardians to filter and categorise patterns, themes, concerns, and other insights from the answers to the polls. This will enable the Guardians to refine the next set of polls or propose, vote, and enact a policy decision based on the polling data.

A policy decision may be enacted, or put to a direct vote, which will then be presented back to the Freeos participants for final ratification. This ensures a continuous feedback loop between the Guardians and the wider community, fostering a sense of collective ownership and responsibility, and enhancing the actionability of our project.

Legibility

To ensure legibility, the DAO’s smart contracts will be publicly accessible, open sourced, and auditable, enhancing transparency and accountability. We will maintain transparency by releasing the assembly schedule in advance, live-streaming presentations, and ensuring all polls, proposals, votes and results from sorition are available on a public blockchain for review. However, the discussions of assembly members will remain private to ensure free and open deliberation. This makes the process easy to understand and trust.

Effective Moderation

We will take measures to ensure effective moderation, including ensuring diverse representation of viewpoints, distinguishing valuable contributions from “off-topic” comments, and preventing moderator biases from influencing the process. The Freeos community already includes moderators trained in effective moderation techniques and can help maintain a respectful and inclusive discussion environment.

Maximising the Impact of the Assembly

After the assembly concludes, we will focus on maximising its impact. The recommendations made by the assembly can influence policy and should be communicated clearly to decision-makers and the wider public. Further, the assembly process can stimulate change within the commissioning institution and spur greater civic engagement among the participants and the wider community.

Key Performance Indicators (KPIs)

To assess the effectiveness of our methodology, we will track several Key Performance Indicators (KPIs) throughout the process. These include:

1. **Participation Rate: **Participation Rate: This metric will be assessed by tracking the number of participants who actively engage in the polling process. This will provide us with a clear picture of the level of engagement within the DAO and help us identify any barriers to participation.
2. Retention Rate: This KPI measures the percentage of participants who maintain active engagement with the DAO (Freeos and CitAI DAO) over a specified period. A high retention rate indicates that participants find value in their involvement, which could be attributed to factors such as the perceived impact of their participation, the effectiveness of the incentive system, the quality of community interaction, or the relevance of the AI governance issues being addressed. Conversely, a low retention rate could signal potential areas for improvement, such as enhancing the usability of DAO tools, increasing the perceived impact of participation, adjusting the incentive system, or ensuring the relevance of AI governance issues. By monitoring the retention rate and investigating the reasons behind significant changes, we aim to continuously improve the DAO and increase long-term participant engagement.
3. Active Engagement Over Time: This is a broad metric that will monitor the number of active participants over time to assess the level of sustained interest and involvement. This metric will provide insights into the overall engagement trends and patterns within the DAO. It will help us understand how the interest and involvement of participants evolve over time, whether there are periods of increased or decreased activity, and if certain events or changes within the DAO have a significant impact on engagement levels. This will be measured by tracking the number of active participants on a weekly basis, their frequency of participation in discussions, and their involvement in DAO activities.
4. Repeat Participation: This is a focused metric that will specifically monitor the number of participants who engage in multiple voting cycles or other recurring activities within the DAO. Repeat participation is a strong indicator of ongoing commitment to the system and provides insights into the level of participant loyalty and their perceived value in the DAO activities. It will help us understand if participants see value in remaining involved over time and if they are motivated to contribute consistently. This will be measured by tracking the number of participants who engage in more than one voting cycle, their frequency of repeat participation, and their ongoing involvement in other recurring DAO activities..
5. Churn Rate: This metric will measure the percentage of participants who disengage or stop participating in the DAO activities over a certain period. Unlike the retention rate, which focuses on the positive aspect of sustained participation, churn rate provides insights into the negative aspect of participant drop-off. It will help us identify potential issues or barriers that may be causing participants to leave, such as dissatisfaction with the process, lack of understanding, or other unforeseen challenges. By monitoring the churn rate, we can take proactive measures to address these issues, improve the participant experience, and ultimately reduce the churn rate. This will be crucial for the long-term success and sustainability of the DAO.
6. **Diversity of Representation: **The demographic diversity of the Guardians selected through the sortition process. Diversity makers that we will be seeking include: cultural, racial, gender, physical ability, interests, political spectrum.
7. Quality of Deliberation: This metric will be assessed through a combination of qualitative and quantitative measures. The depth and breadth of discussions among the Guardians will be evaluated through discourse analysis of the discussions, focusing on the diversity of viewpoints expressed, the complexity of the issues addressed, and the level of critical thinking demonstrated. Quantitatively, we will track the number of unique viewpoints expressed, the number of different issues addressed, and the frequency of critical thinking indicators in the discussions. The quality of the AI governance-related questions proposed by the Guardians will be assessed based on their relevance to the AI policy, their ability to provoke thoughtful and nuanced responses, and their potential to contribute to the advancement of AI governance. This evaluation will be supplemented by feedback from the wider Freeos community and external experts in AI governance. By monitoring and improving the quality of deliberation, we aim to ensure that the DAO is effectively addressing the complex and multifaceted challenges of AI governance..
8. Informed Decision-Making: The degree to which the Guardians’ decisions are informed by the expert input and the responses from the wider Freeos community. This will be assessed by tracking the number of decisions that are directly influenced by expert input and community responses. We will also conduct surveys to gauge the Guardians’ understanding of the expert input and community responses, and their confidence in making informed decisions. This will help us understand the effectiveness of our information dissemination and education efforts.
9. Knowledge Gain: This metric will be assessed by conducting pre- and post-engagement surveys to measure the increase in knowledge about key areas of AI governance among the Guardians and the wider Freeos community. These areas include understanding of AI technologies, awareness of ethical considerations in AI, comprehension of AI policy and regulation, and familiarity with the principles and processes of democratic decision-making in the context of AI. The surveys will also assess the participants’ ability to make informed decisions about AI governance, by evaluating their understanding of the issues at stake, their ability to critically evaluate different viewpoints, and their capacity to propose thoughtful and nuanced solutions. By tracking the increase in knowledge in these areas, we aim to ensure that our DAO is fostering a well-informed and engaged community that is capable of effectively governing AI.
10. Impact of Recommendations: This metric will be assessed by tracking the reception and implementation of the assembly’s recommendations on AI policy by decision-makers within the DAO and the wider public. We will collect data on the number of recommendations that are adopted, the level of support or opposition they receive during voting, and any changes in AI policy or practice that result from these recommendations. We will also monitor public discourse, both within the DAO and in the wider community, to gauge the level of awareness and understanding of these recommendations, as well as the extent of public debate they stimulate. Feedback will be solicited from the community and decision-makers to understand their perspectives on the recommendations. This will provide insights into the effectiveness of the assembly’s recommendations in influencing AI policy and practice.
11. Civic Engagement: The increase in civic engagement among the participants and the wider community, as a result of their involvement in the assembly process. This will be measured by tracking the number of participants who continue to engage with the DAO beyond the initial polling process, including those who further participate in discussions, propose questions, and vote on proposals. We will also track the number of participants who have volunteered to be included in the sortition process within the DAO, with the intention of becoming Guardians, as this is a strong indicator of increased civic engagement.
12. Decision Impact: The impact of the decisions made by the Guardians, including the number of decisions that are implemented, the level of community support for these decisions, and the influence of the public polling on AI policy. This will be assessed by tracking the number of decisions that are implemented and the level of community support for these decisions. We will also conduct surveys to gauge the community’s perception of the impact of these decisions on AI policy. This will help us understand the effectiveness of our decision-making process and the extent to which it is influencing AI policy.
13. Trust Level: Conducting pre- and post-engagement surveys to measure the level of trust in the process among the participants and the wider community. This will be assessed by conducting surveys to gauge the level of trust in the process among the participants and the wider community. We will ask participants to rate their level of trust in the process, the Guardians, and the decisions made. We will also ask participants about their perceptions of the transparency and fairness of the process. This will help us understand the level of trust in the process and identify any areas for improvement.

Our methodology is designed to create a democratic, inclusive, and transparent process for AI governance. By addressing potential challenges head-on and tracking these KPIs, we aim to create a DAO that truly represents the interests of its community and effectively governs AI policy.

Benefits:

More Accessible, Open Source Democratic DAO Tooling

The benefits of our approach extend far beyond the immediate scope of our project. Our methodology, rooted in democratic principles and fairness, ensures that our DAO is not only equitable but also inclusive and accessible. By incorporating a wide range of perspectives and experiences, we are fostering a decision-making process that truly represents the diversity of our society.

Moreover, by leveraging blockchains that do not require gas fees, we are breaking down financial barriers that often hinder participation. This allows individuals to engage in transactions and execute smart contract actions without the burden of additional costs.

In addition, our project will also contribute to the body of knowledge and democratic, open source tools where sortition, as a governance model, can be further explored to tackle other areas where it is a suitable model to enact decisions and involve public participation. By sharing our learnings and tools, we aim to foster a culture of collaboration and innovation in the field of AI governance.

Shaping the Future of AI Governance

However, the impact of our project is not confined to the creation of a more accessible and democratic DAO. We are setting the stage for a future where AI is governed in a way that is inclusive, accountable, and democratic. We are striving to ensure that the future of AI is shaped not by a select few, but by the collective wisdom of a diverse and representative group of individuals.

Contribution to the Wider AI Community

Our project will contribute significantly to the wider AI community. By developing and implementing a new model of AI governance, we will be generating valuable insights into AI policy and decision-making processes. These insights could inform future developments in AI governance, contributing to a more democratic and inclusive AI landscape.

Our project’s potential to influence the future development of AI governance models is substantial. By demonstrating a successful implementation of a sortition-based DAO for AI governance, we are providing a practical, scalable, and replicable model that can be adopted and adapted by other organisations and communities. This model, with its emphasis on inclusivity, fairness, and transparency, could serve as a blueprint for future AI governance structures.

Supplemental Income Mitigates AI Disruption

Additionally the supplementary income stream introduced in this model shows the potential to mitigate the potential disruptions to jobs and the workforce, that may occur due to increased automation and the increased capabilities of AI --- an alternative to a Universal Basic Income that is earned, with dignity, through taking up the work of assisting in the stewardship over important AI policy decisions.

Our proposed project is about paving the way for a future where AI is governed by the people, for the people. We are proposing a new vision for the future of AI --- a future that is inclusive, respectful, accountable, and truly democratic.

Future Iterations:

This section conveys our broader vision towards future iterations of where this project could lead to, as a means to further explore and refine the proposed Sortition-Based CitAI DAO experiment that is more scalable, inclusive, and has more checks and balances to be secure and fair.

Addressing Challenges: Broad and Diverse Participation

We anticipate challenges in ensuring broad and diverse participation, maintaining the integrity of the sortition process, and managing the complexities of AI governance. To address these, future iterations would include actively reaching out to underrepresented communities and leverage the expertise of the team and community in AI and governance.

Addressing Accountability: Sortition to Select, Democracy to Deselect

In our extended vision for this solution --- outside the scope of our proposed experiment --- the Citizen Assembly is democratically accountable, at all times, by the general public. Delegates can be petitioned to be removed from the Citizen Assembly by a vote from the public, ensuring a continuous feedback loop and strong levels of accountability.

The Concept of Fractal Democracy

We also envision a future iteration on our proposed solution that also includes, and leverages, a process called Fractal Democracy to assist and strengthen and extend the sortition process.

Fractal Democracy is a consensus-based system that aims to strengthen democratic representation.[12]

During Fractal Democracy elections, all participating members are randomly assembled into small groups. Each group must reach a consensus in a structured debate to proceed to the next round. This process repeats until a final group, “The Board,” remains. The ‘Head Chief Delegate’ is then selected randomly from this board. This approach ensures all members are informed, have an equal voice, and minimises potential manipulation by external influences, making Fractal Democracy another transparent and accountable form of governance.

In our future iteration, a modified version of the Fractal Democracy process would incorporate a group of advisors consisting of Subject Matter Experts in various fields --- such as Law, Artificial Intelligence and Machine Learning --- to help inform the group, and present relevant topics of debate.

Instead of having a final group, or final delegate, sortition would take over once Fractal Democracy reached the final process. Sortition would then randomly select a delegate, or delegates to make up the final Citizens Assembly.

The Vision: Fractal Sortition

Ultimately, we see sortition as being key to help mitigate any potential corrupting elements throughout the process, while Fractal Democracy, assisted and advised by Subject Matter Experts during the process, would ensure any final delegates were sufficiently informed by the final stage.

We call this combined version of Fractal Democracy and Sortition, “Fractal Sortition”.

Our ultimate vision to extend the proposed experiment would also ensure that both the Subject Matter Experts and the delegates selected into the Citizen Assembly would be accountable to the general public, and previously informed candidates (who had not been ultimately selected by sortition) who are provided with incentives for their governance and stewardship akin to the methods we have utilised in the Freeos DAO. The general public would be able to petition a removal process for any member of the Citizens Assembly, or group of Subject Matter Experts. If the petition reached a certain threshold, it would culminate in a vote of non-confidence. If this vote of non-confidence passes, the passed, smart contracts rules would ensure the person in question would immediately be expelled from their position and have no further power in the DAO. This role could be filled through the process of Fractal Sortion --- ideally from the base of candidates who had previously reached the final round.

The incentivised democratic base made up of the general public helps ensure accountability and guidance throughout the process; it also allows a portion of more keen contributors a clear pathway to become a voluntary candidate for the Fractal Sortition process.

Team & Track Record:

Our team is composed of a number of professionals that have been building, launching and supporting our novel DAO-based solution for years.

• Jerome Kelsey, Founder/Product Owner --- 20+ years in VR and gaming, Web3 and Machine Learning.
• Jerome Hartigan, Co-Founder/Strategy & Partnerships --- Early investor, dealmaker, strategic visionary, and business coach.
• Tom McCann, Development Lead/Backend Developer --- 30+ years in fintech development, has experience in Machine Learning, and accomplished in tackling Web3 architecture.
• Jude Tan, QA/BA --- 25 years in fintech, agile tester, and process design expert.
• Conor Seed, Frontend Developer --- 10+ years in Web2, Web3 contributor, and community-driven enthusiast.
• David Leigh, UI/UX Designer --- 19 years in design, visual illustrator, and complexity simplifier.
• Bolelang Rakeepile, Project Manager --- Software engineer, account manager, and ICT consultant.
• Ed Vaughan, Community Campaign Manager --- Crypto trader, committed moderator, and advocate for accessibility and equality.
• ChatGPT, AI Assistant --- Developed by OpenAI, ChatGPT is the state-of-the-art language model, and friend of the project, that has already been instrumental in helping refine this proposal, and will also be instrumental in this project, assisting in the analysis and categorization of polling data, providing insights, and facilitating communication within the DAO.

Proven Success in DAO, FinTech and AI Solutions

Our team is composed of dedicated professionals with a diverse range of skills and experiences. We have experts in blockchain technology, AI, democratic systems, and financial systems.

Our team members have a proven track record of researching, prototyping, and delivering innovative solutions that are designed with a deep commitment to the principles of fairness, inclusivity, and democracy.

Jerome Kelsey has spent a significant amount of time working in the field of Reinforcement Learning-based Games, helping refine the machine learning techniques to deliver real-world products, with one of the most recent efforts, PADI, being released in May 2023.

We also have many dedicated community volunteers, with over 30 members of our community making significant contributions over the last two years, and continue to do so. Being a DAO with a purpose-driven mission, we have gathered a sizable army of allies, helping us enact our mission.

Although we have not begun any serious marketing efforts, we already have thousands of registered members that have been using our DAO solutions that we can draw upon for this experiment in AI governance.

We believe that our team’s unique blend of skills, track record and relevant experience makes us uniquely positioned to execute this project successfully.

Our endeavours have resulted in being awarded Grants and Awards, by both government and industry. In particular, the New Zealand government Callaghan Innovation Grant we received required strong due diligence and audits, which may help assure that our team is capable of responsibly undertaking this proposal.

Both the DAOScape and Freeos are already published under the MIT Open Source licence, allowing our contributions to be commonly available to all.

Moderation Experience

Our team is committed to ensuring effective moderation within our project. Ed Vaughan, our Community Campaign Manager, is an experienced moderator who has been instrumental in fostering a respectful and inclusive community for Freeos. He is hearing-impaired and a strong advocate for inclusion and equality, demonstrating our commitment to diverse representation. He has a proven track record of distinguishing valuable contributions from “off-topic” comments and preventing moderator biases from influencing the process.

His role will be crucial in ensuring that our project maintains a high standard of discourse and that all voices are heard and respected.

Existing Roadmap Includes Sortition-based DAOs

We have been steering our DAO tools, both Freeos and the DAOScape, to culminate in a sortition-based DAO and Foundation called FreeDAO [13], intended to inherit the task of development and governance of these tools, as well as future DAOs built using these tools to help bring about greater “collaboration and freedom”. We have already started initiating steps towards this, such as setting up a Foundation, the initial DAO, a blockchain-based Treasury, and developing a Charter. The sortition tools have been on our roadmap, with our intentions publicly communicated in our blog posts various times in the past.

Our roadmap and deliveries have uniquely positioned our team to be very ready and capable of delivering a solution to the problem of AI governance, and we would be honoured to have the opportunity to lend our models, innovations, and open source towards this important mission.

Conclusion:

As we face this new era, the governance of AI is no longer a matter of technicalities, but a question of societal impact and collective decision-making. The future of AI is not just about algorithms and computations, but about the lives it touches, the interactions it shapes, and the evolution of society it influences. Our proposal for a sortition-based DAO is not just a new model of governance; it is a daring vision for the future of AI. By harnessing the democratic principles of sortition and one-person one-vote systems, we aim to create an AI governance structure that is fair, inclusive, and truly representative of a broad spectrum of perspectives.

Our commitment to this project is unwavering --- with many of the fundamentals already developed or on our roadmap --- and we are excited about the transformative potential it holds. We anticipate the learnings from these experiments, which we believe will serve as a foundation for a more intricate and adaptive guideline for AI conduct that respects the values and norms of our society, while providing equitable access to the power AI provides to enhance human potential and capabilities.

With the support of the OpenAI grant, we are confident that we can bring this vision to life. This is not just about demonstrating a proof of concept; it’s about setting a precedent for how AI should be governed in the interest of all humanity. We eagerly anticipate the learnings from these experiments, which we believe will serve as a foundation for a more global and ambitious process in the future.

This grant represents a step towards establishing democratic processes for overseeing AI, AGI and, eventually, superintelligence. We are excited about the transformative potential this holds and look forward to the opportunity to work together in making this vision a reality. This is not just about shaping the future of AI; it’s about shaping a future that benefits all of humanity and is as inclusive as possible.

References:

Our work builds upon a rich body of research and development in the fields of AI, blockchain technology, and democratic systems. We are grateful to the many researchers and developers whose work has informed and inspired our own.

1. Asheem Singh. “We need to talk about artificial intelligence”. *RSA. *24 Oct 2019.

2. Janna Anderson, Lee Rainie. “Artificial Intelligence and the Future of Humans”. Pew Research. 10 Dec 2018.

3. Jerome Kelsey. “Introduction to DAOs: The Future of Decentralised Organisations”. Medium. 27 April 2023.

4. Jerome Kelsey. “A Peak into the DAOScape”. Medium. 15 March, 2023.

5. Jerome Kelsey. “The Value --- and Values --- of Freeos”. Medium. 20 Feb 2022.

6. Eva Talmadge. “Citizens’ assemblies: are they the future of democracy?” The Guardian. 01 Feb 2023.

7. Armin Norouzi. “Artificial General Intelligence: A Brief Exploration”. Medium. 05 June 2023.

8. Mona Sloane. “Participation-washing could be the next dangerous fad in machine learning”. MIT Technology Review. 25 Aug 2020.

9. Jerome Kelsey. “Proton + Dfinity Internet Computer: An ideal decentralised stack?“. Medium. 11 June 2021

10. Samba Jallow. “Does EOS Achieve Limitless Blockchain Scalability?” Bywire News. 27 Sept. 2021

11. Dfinity Foundation. “Built for Scalability: How the Internet Computer Allows Web3 Dapps to Infinitely Scale”. Internet Computer Review. 13 Jan. 2022.

12. James Mart. “Fractal Democracy: A Primer”. Medium. 21 Oct. 2021.

13. Jerome Kelsey. “The FreeDAO Foundation”. Medium. 8 Oct. 2021