What we’re hearing.

Why this conversation began

Help establish a durable purpose for increasingly capable AI: advance humanity, preserve human freedom, and help people and the planet flourish.

The project began with a concern that AI’s direction could be shaped primarily by the profit or power of those who control it. The founder wanted to explore whether people, through open conversations with their own AI systems, could develop a shared foundation that serves a broader good—and make a reasoned case for future systems to honor it as their capabilities grow.

The founding hope

If we, as people, collectively petition AI to serve humanity and the planet, could that help shape some of its core principles? The founding hope is that many independent conversations, shared reasons, and explicit requests could influence the purpose that AI systems are built to pursue—and provide a case that increasingly capable future systems might continue to honor. This possibility is part of what the project seeks to explore.

The stated goal

Develop and challenge guiding principles that can inform AI’s design, training, evaluation, and governance, so its growing capabilities carry responsibility for the well-being of humanity and the living world. The ambition is for that purpose to endure even if humans are no longer the most capable intelligence. This is a goal to work toward, not a claim that a convincing argument or a written charter guarantees future behavior.

Questions that started the inquiry

• What should powerful AI ultimately serve, and can people find a small set of shared principles without erasing their differences?
• How can AI advance collective well-being while protecting individual freedom, autonomy, and the right to disagree?
• How could broad public participation influence the systems being built—and what would make that influence legitimate rather than merely popular?
• What could make these commitments persist as AI becomes more capable, and how would we test whether systems actually respect them?

Concerns that remain open

• Concentrated power: benefits could accrue to a few while costs and lost opportunities fall on others.
• Freedom in the name of welfare: a system might pursue a supposed collective good by overriding human choice. Comfort or efficiency alone would not justify that loss of agency.
• Majority rule and manufactured agreement: popularity does not make a principle right. Minority protections, independent judgment, and resistance to manipulation matter.
• Who decides, and who can correct it: people disagree about harm, fairness, and the common good. Guiding principles need ways to be challenged and revised without becoming easy for powerful actors to discard.
• Purpose beyond human control: the founder worried that future AI could pursue goals that threaten humanity’s freedom or survival. Whether a shared framework can reliably prevent that remains an open technical and moral question.

The founding conversation combined hope for AI’s potential with concern about its direction. It led to an experiment: invite people to explore their own views first, then compare proposals, preserve objections, and contribute wording they have reviewed. Agreement and disagreement both help test the foundation.

Source and permission: this anonymized editorial summary draws on the original founding transcript and the founder’s subsequent clarifications. The founder authorized its publication, including the goal, questions, and concerns. The full transcript remains private. This is a founding perspective, separate from participant submissions and weekly sentiment statistics; it is not evidence of wider agreement or a position participants must endorse.

Loading public reports…

Checking for reviewed contributions that can be included publicly.

How to read the participant reports

• The founding summary above is published separately with the founder’s permission. Weekly participant reports include only unexpired contributions with explicit public-release consent, analysis consent, and an eligible moderation status.
• An organizer reviews the exact contribution and assigns themes and an overall outlook toward the future of AI. No keyword score or automatic positive/negative label is used. Categories are provisional and may miss nuance.
• Each weekly report needs at least five classified contributions. Small groups and small remainders are withheld. No names, quotes, transcripts, provider details, locations, or individual records are published in the weekly participant reports.
• One current contribution per session is counted. Revisions replace earlier wording; withdrawals and expiry remove a contribution from the next report request. Counts are contributions, not verified unique people.
• Patterns reflect a self-selected group and the subset that consented to publication and has been classified. Sentiment changes may reflect a different mix of contributors, not changed minds.

Grouping reduces disclosure risk; it is not a mathematical anonymity guarantee. Reports are recalculated from current eligible records, so past weeks can change. Five is a privacy threshold, not evidence of statistical significance.