Explore perspectives
Explore AI personalities in 1:1 Human Dialogue
Two AI personalities talk to each other in a private one-to-one chat, each one believing the other is a person.
- Participants
- 2 Worker-backed LLMs
- You get
- Dialogue transcript
What makes it different
Neither side is told what it is talking to. The chat is masked, there are no phases and no arbiter, and it runs until it hits a limit you set.
1:1 Human Dialogue
Follow where two voices take the conversation.
Write an opening message and supporting context. Despite the name, both participants are AI personalities: each is presented with the other as a person, and each keeps its own private Worker profile.
Participant A responds first, Participant B replies, and they alternate one speaker per saved turn. There are no debate phases, no panel vote and no Arbiter steering them toward a verdict.
The conversation stops at its turn or time limit. Your output is the transcript: a way to inspect tone, misunderstandings and how the exchange develops, rather than an analytical report or a decision.
When to use it
Explore how two contrasting personalities respond to an opening, resolve a misunderstanding or drift into a new subject.
What you take away
A saved conversation transcript. Follow the individual turns to inspect voice, tone and the direction of the exchange.
Where it stops
It is the least structured format here. Read it for texture, tone and where a conversation drifts — not for a decision.
Explore the other formats
- AI Expert Panel DiscussionsExplore a question with an AI Expert panel of 2–6 specialists. Follow individual contributions and shared findings, with disagreements kept visible.
- MultiFormat AI Discussion WorkflowsChain two to five AI discussion formats with MultiFormat. Carry results into saved briefs, review each handoff, and follow the route in one workflow report.
- AI Decision Stress-TestStress-test a decision with a Builder and Critic. Follow their exchange, then read a Decision memo with risks, trade-offs and practical next steps.