Mosaic is a non-profit R&D lab working to help groups coordinate, reason effectively, and resolve conflict. We are building interfaces which support new kinds of collaborative discussions, promote overlooked perspectives, and encourage nuanced exploration of conflicting viewpoints.
Rapid developments in AI are quickly transforming the future of reasoning, communication, and conflict. Our goal is to ambitiously apply AI to collective deliberation problems, explore the frontier of human-machine collaboration, and to keep human reasoning at the center of important decision making.
Differential Development
Technological progress does not guarantee we use new technology responsibly, nor that we manage to achieve cooperative and mutually beneficial outcomes.
To safely navigate our future, we need systems which directly enable coordination, empower human groups to think clearly, and prevent disastrous conflicts. The order we develop technology matters. By accelerating work which uses recent advances in AI capabilities to enable better collective reasoning and coordination, we aim to give people the ability to navigate an uncertain future.
Human-centered Design
AI systems are improving rapidly. For this reason, it is important that work to augment human reasoning improves alongside AI capability advancements, and is not replaced or made obsolete by them.
We are especially concerned about the impulse to outsource reasoning to AI systems, and to fully replace humans in the deliberation process. In order to keep humans at the center, we are working to develop and demonstrate methods for using AI mediation to support and enhance human-lead discussions.
Code as a Catalyst
By testing and demonstrating new methods for AI-assisted deliberation, we aim to help support a movement directed toward improving human reasoning, large scale collaboration, and coordinated action.
We believe software tools can serve as a catalyst for social change by both demonstrating what is possible, as well as becoming a part of the language we use to imagine the future. Our goal is to advance the discussion about AI's role in society by building software which concretely embodies our ideas.
Projects
Nexus: AI-mediated deliberation
Built with support from Lightspeed Grants
Nexus is a work-in-progress, experimental platform for AI-mediated group deliberation. It uses large language models to help groups understand each other better and surface insights and solutions that wouldn't have been found otherwise.
Pantheon: AI-facilitated metacognition
Built with support from the Long-Term Future Fund
Pantheon is an experimental AI-facilitated metacognition tool and research prototype exploring new ways to use LLMs to improve human thinking. Intricate, customizable AI characters and a novel, tree-like UI structure empower the user to think and work in new ways.
Team
