Step 1
Start with one question and two people.
V1 begins with a deliberately small unit: one shared dilemma, one invite link, and one timed conversation.
Deliberate
Research Lab
Structured human deliberation
Deliberate
A lab exploring how AI can help people deliberate more clearly today, and how those deliberative outputs might shape the systems that increasingly act on our behalf.
What we're testing
Deliberate is a research program, not a claim that the whole stack already exists. V1 is the first instrument.
Different ways AI can step in
We are testing different ways Seren can intervene in a conversation, from clarifying terms to surfacing the crux to offering a light synthesis when it is warranted.
Whether those interventions reliably improve deliberation
We want to know whether some forms of AI guidance can consistently make deliberation clearer, more focused, and more useful without taking over or thinking for participants.
Whether deliberation can become input for agents
We are also exploring whether the outputs of human deliberation can become useful goals and constraints for agents that negotiate or coordinate on our behalf.
Whether Seren can become a real-time facilitation agent
Beyond V1, we want to explore whether Seren can become a callable agent that joins live conversations, works across formats, and connects to other platforms when clearer deliberation is needed.
How it works
The current system is intentionally narrow: a structured, dyadic deliberation flow designed to make disagreement easier to see and discuss.
In the future, we expect to test real-time and multi-participant formats as well.
Step 2
Use AI guidance sparingly.
Seren intervenes only when a short clarification or sharpening question can make the real disagreement more legible.
Step 3
End with a more explicit record.
Participants leave with reflections and a debrief that capture what mattered, what shifted, and what stayed unresolved.
Why it matters
The stakes are broader than one experiment. We care about how people deliberate together now, what that means for democratic capacity, and whether better collective reasoning can become a useful input to the systems surrounding us.
Democratic capacity
Technology should strengthen our capacity to reason together.
In a democracy, AI poses a serious risk if it weakens people's ability to deliberate, judge, and make sense of shared problems with one another.
A better direction is technology that enhances those capacities instead of bypassing them, especially when the stakes are collective and the disagreement is real.
Join the work
Try the current experiment, or join the Deliberate list.
V1 is live now as a small-scale deliberation instrument. If you want results, relevant essays, and future experiment invitations, you can sign up here.
Get Deliberate updates
Results, relevant essays, and future experiment invitations.