This website is a work of speculative fiction. Built as part of Stuart Candy's Spring 2021 Experiential Futures course at CMU.
Introduction - Who is SIMONE
The 2032 Year-in-Review breaks down the roll-out of two processes — causal discovery AI (SIMONE) and adaptive policy-making. Causal discovery AI is artificial intelligence that claims to move beyond correlation to highlight the precise relationships between cause and effects. Adaptive policy-making claims to be a generative process in which underserved communities, including grassroots organizations, play a large part in prioritization.
These two efforts run parallel, and are managed by the volunteer-run organization Wikipedia’s new policy division called Wikipolicy. Wikipedia is part of a coalition of technology companies, which includes Facebook, Amazon, Apple, Netflix, and Google (also known as FAANG), that was birthed from the 2021 Congressional hearings.
Wikipolicy’s objective is to add emphasis on accountability, transparency, and logics of care in algorithmic and participatory policy-making’s advancements. They claim that they do this with a consciousness that is hyperaware of the critiques and accounts of biases with AI’s inception.
Background
In 2021, Congress held a hearing to tackle the claim that big tech lacked accountability and were at the helm of the phenomenon of disinformation. Congressional hearings that questioned the methods and approaches of social technology platforms are not new; our key actors in “big tech” — Facebook Inc., Twitter Inc., Alphabet Inc. and its Google unit — have been meeting face-to-face with Congressional representatives since 2017.
2021 was, however, a watershed moment.
Four key results from the hearing:
Big tech companies are regulated in that triple bottom lines are mandated; watchdogs are developed; whistleblowers are protected; mental health is a key component; and contracts with DARPA / US military are to cede.
A coalition of technology companies focuses on the development of BOTH causal discovery in AI and generative (and kind) policy-making.
Participatory policy-making extensively expands the inclusion of grassroots level organizations, such as mutual aid groups, community based organizations, and student-run organizations/groups.
Accountability team is a public and private partnership. This further unites Republican + Democratic parties without perpetuating universalism.
For nearly a decade, the conglomeration of big to small tech companies collaborated in pods to test small scale prototypes while chipping away at large scale concepts. The purpose of this strategy was to cover as much ground as possible while meeting the primary benchmark: kick-off the roll-out plan by January 2032.
And, this was largely successful.
The plain entailed:
Trust-building between public and technology companies (Q1)
Trust-building between public and private sectors (Q1)
Building grassroots network to inform politics (Q2)
Make changes to policy-making processes (Q3)
Build out the causal discovery AI and adaptive participatory policy-making processes (Q4)
2032 Year-in-Review
Quarter One: Trust-Building
The first quarter focused exclusively on laying a strong foundation of trust across two key relational lines — the general public and technology companies; and the public sector and private sector. All parties understood that embodiment of key principles was crucial to setting up this effort — the launch of causal discovery AI and generative policy-making — for success. Each approach aimed to instill the key principles of transparency, explainability, and care; this was applied to the dynamics across the four key stakeholders mentioned.
Process One: In Full Candor
In Full Candor focuses on two key principles: transparency and explainability. To further disclose the undercurrents they uniquely address: to be transparent makes explicit the logics that underpin algorithmic processes; to be explainable makes clear the reason for any output. These two principles are underscored so that the development of both AI and policy-making may be generative in nature.
In Full Candor finds inspiration from the Fish-Bowl Reconciliation method. This approach to reconciliation is meant to “ventilate” hyper sensitive topics or sharing of ideas. The intent is to bring transparency to decision making processes and to rapidly build trust. The general idea is that rather than a large group having an open discussion about something, which can be difficult to handle and often only benefits a few active participants, a smaller group (ideally 3 – 6 people) is isolated to discuss while the rest of the participants (maximum of 50 people) sit around the outside and observe without interrupting. Facilitation is focused on the core group discussion.
In these discussions, participants were expected to explicitly share the logics underpinning an algorithmic process and the reason for outputs. The explicitly and clearly stated items were then thoughtfully and rigorously discussed within this fish-bowl context.
This became the hallmark of fostering a culture of respect, transparency, and trust while embedding accountability, building knowledge, and instilling care.
Process Two: A decentralized ecology of radical care
The decentralized ecology of radical care focused on trust and teachability. Its makeup rooted itself in the dimensions of feeling and/or affect. The key stakeholders were equipped with the ability to articulate and read the affective qualities of knowing, being, and doing. These included feeling, language, and energy.
Feeling reflects the degree of attachment and depth of one’s optimism regarding an issue, circumstance, and relationship.
Language reflects the contours of a budding culture of a locale, organization, or close-knit community.
Energy reflects the potential for deeper trust building or waning.
Initially skeptics kept the group from moving forward as indicated in the first month of applying such metrics. However, upon gaining buy-in, the two parties saw the visualization of behavioral patterns that forecasted a futurity of respect, transparency, and trust. Similar to the Algorithmic League of Justice.
Trends
Respect, transparency, and trust
Deep embodiment of the principles of motivations in the development of two milestone achievements (causal discovery AI and generative policy-making)
Key Stakeholders
The general public
Technology companies
Government
Private sector
Key highlights
Dimensions of trust and accountability: inter-relationally & algorthimically
Transparency & explainability
Affect & emotion
Fish-bowl effect
Quarter Two: Community Engagement
As the first quarter came to a close and a system of trust was established between the public and private sectors, and the public and technology companies, there was ample space for the establishment of grassroots networks and relationships across the country in order to set the stage for a more holistic and representative policy making process.
This network was built off the back of the model of mutual aid organizations in an effort to continue to follow the trends of transparency established previously, and work to include non-hierarchical data and opinion collection in localized communities. One of the primary challenges faced in this process was the legitimization of the framework in the eyes of some parties, as the empowerment of informal and ‘non-expert’ opinions was a new concept to many.
Process One: Identification
In order to gain a sense of connection between the local parties, it became imperative to outline an understanding of what issues were most pressing to each of these communities and help to build a sense of empathy within them. Wikipolicy worked to recreate strong intra-community ties as the general public moves further from the importance of ‘place’ as a sense of identity as the availability and readiness of remote work and ease of long-distance connection lessens the import of location.
Horizontal open discussion meetings allowed space for community members to come forward with issues of personal importance to a set topic of discussion. Allowing only a limited number of people in a given meeting to discuss an issue keeps the discussion focused, and helps to set the basis of a hub for community opinion and organization. Identifying a member of the local municipality allowed for easier organization and communication within the larger group. Underlying these discussions is the importance of openness, respect, and welcoming as a baseline for entry.
Process One: Co-Creation
While forming a series of smaller, more isolated networks within a larger group may not be a new experience for many participants, the process of self-empowerment through voicing their individual expertise may not be as familiar to them. Our desire through this process is to treat each individual as an expert of their own experience and bring that knowledge to the discussion table as a valid and important perspective for creating and understanding policy moving forward.
This is a change from the traditional top-down structure in which policies are determined and created by elected representatives. The importance of policies and the goals targeted by each are now determined by the community and the grassroots groups that arise out of these meetings. In order to build a sense of confidence for these new activists, Wikipolicy implemented a series of co-design and co-creation training sessions to provide an understanding of the processes involved in organizing and communicating the necessary steps to building support for ideas.
Trends
Respect
Empowerment
Education
Organization
Discovery
Key Stakeholders
Localized communities
New activists
Key highlights
Empowerment
Co-creation
Community-building
Empathy
Quarter Three: Causal Discovery AI Roll-Out
In the third quarter, policy creation ceased to be a top-down process. The power to bring forth new policy shifted from the legislators to the people. The coalition viewed this transition as three discrete phases—Aggregation, Translation, and Validation.
Phase I: Aggregation
After the general comment period facilitated by grassroots organizations nationwide, the coalition collected copies of artifacts generated during these discussions. Most of the data was cleaned up by using annotation tools to help the AI to quickly and easily interpret the artifact and draw out ideas.
Phase II: Translation
Once the AI captured the essence of each artifact, it analyzed the set of collected data for patterns across groups, trends in opinions, breadth and depth of individual issues, as well as the various associated costs of each issue.
The output from the analysis contained two components:
The first component was a graph representing a network of interdependencies linking various issues. While elementary, this was viewed as a great achievement because it demonstrated a rudimentary understanding of causal relationships.
The second component was a set of policy recommendations that draws links between the nodes in the social issues graph, accompanied by an explanation of their relationship. Each policy recommendation included considerations such as potential impacts and resource allocation.
Phase III: Validation
To avoid replicating existing biases in any new policy, Wikipolicy partnered with watchdog groups to form a committee to audit both the output from the previous phase. The committee’s goal was to eliminate connections that reflect historically discriminatory policies and practices.
Once properly pruned and edited, both graph and policy recommendations were fed back to the algorithm to produce an updated version of each. This iterative process will be an ongoing effort to ensure the AI behaves in a fair manner.
Trends
Transparency & accountability
Causal discovery
Bias audits
Key Stakeholders
The general public
particularly marginalized communities
Key highlights
Participatory policy creation
Bias checking
Quarter Four: Adaptive Policy-Making Roll-Out
While just at the beginning, the coalition is dedicated to rebuilding the bridges broken in previous years by Black Box AI and partisan government. The technological advancements in casual discovery opened up new ways of using algorithms and AI as methods of transparent government into a participatory policy making system that reflects the will of the people at scale.
Phase IV: Validation
A general election acted as a period for public comments where citizens voted on the policy recommendations. Utilizing the grassroots networks established in previous quarters, participants are informed of the policies generated by the AI. During this period they are able to add commentary, voting, and evaluate priority.
The output from the analysis contained two components:
They are then aggregated, and compiled into legally viable reports and used to inform policy makers.
The full results of the elections served as a blueprint to inform legislators about how they should prioritize the various policies to enact.
In this way, all representatives were held accountable for implementing policies that are in the interest of their constituents rather than being swayed by lobbyists.
Looking back, this year brought a lot of change and reckoning to how a bill becomes a law, and who has the power in that process. The implementation of the coalition's AI put a spotlight on how much power corporate and partisan interests have in what makes it to the signing table. The implementation process was not without its flaws, but Wikipolicy pushed for transparency behind these failures and how they were addressed worked to solidify public trust in the AI.
This fall policies generated by the AI made it to the ballot across the country. The 2032 presidential election season provided an ideal opportunity to inform people of possible policy changes recommended through the grassroots networks. Digital authentication methods provided a secure platform for people to participate in during the open commentary period. Lack of partisan rhetoric helped surface issues on both sides of the fence to build areas of unity across local, state and national governments.
Trends
Participatory policy prioritization
Technological innovation for civic good
Key Stakeholders
The general public
Political activists
Policy makers
Key highlights
Local, State, and National policy change
Government accountability
