A live broadcast format

Disagreement, made visible.

For a century, broadcasting offered two things: the monologue, or the brawl. Deliberative media is a third path — several experts, structurally required to disagree, arguing their way to a conclusion in full view of the audience. This is its first working embodiment.

Position A
The crowd is wise — on aggregated questions it beats the expert poll.
Challenge
Wise where the data is clean. Where the money is, it's a herd.
Evidence
Accuracy holds in politics, collapses in noise. Here is the split.
Verdict · conf. 8/10 Not who is right — which assumptions the conclusion rests on, and which of them are fragile.
The form

Not a monologue. Not a brawl. The visible work of thought.

Both older forms leave the viewer outside — handed either a finished conclusion or pure noise, but almost never the process by which a well-founded judgement is formed. Deliberative media puts that process on air: here are the positions, here is the data, here is where the argument is sound and where it is fragile, here is the conclusion and the confidence behind it.

Its purpose is not to replace journalists, but to make the process of weighing the truth visible. The audience does not merely receive a conclusion — it sees how the conclusion was reached, and learns the skill itself.

Until now this format was expensive and rare. Real deliberation demands smart, prepared, genuinely different participants who will neither collapse into chaos nor dissolve into polite agreement. Deliberative media makes that form reproducible.

What defines it

Three things — none of them about technology.

i

Visible disagreement

Different positions collide on air by design, rather than being smoothed into agreement. Disagreement is not a malfunction or a ratings stunt — it is the material itself. It is in the collision of well-grounded positions that the viewer sees where the real fault line lies.

ii

Fact-checking as a live participant

Verification is built into the conversation, not exiled to a caption a day later. An error is corrected in the moment it is spoken — by name, with a source — while the audience is still listening. Caught where it was uttered, not where no one can connect it back.

iii

A map of assumptions, not a verdict from above

A segment ends not with a conclusion handed down, but with the assumptions it rests on laid bare: which are solid, which are fragile. The viewer leaves with an understanding of what an opinion stands on, and where it might collapse.

Why now
In May 2026 the BBC put an AI panel on air. Viewers rejected it — not for being artificial, but for having no real disagreement. The appetite for honest deliberation exists. A format that delivers it did not.
The gap this format is built to fill
The first stage

A working platform, not a concept.

The first embodiment of deliberative media is a live platform, working-titled AI Board Live, online at aiboard.live. The role of structured, differently-tuned experts is played by AI participants under the control of a live human host — they hear human speech, argue along assigned roles, test claims, and reach a verdict, addressing one another by name.

Working today
  • Live human host or AI host — both modes
  • One to seven AI experts per debate, each with a role
  • Hears human speech: voice → text → shared context
  • Follow-up to all, or to one named expert
  • English and Russian, more can be added
Building next
  • Several live human experts in studio, heard by the AI
  • Per-show memory — an evolving cast, held to its past
  • Data connectors so newsrooms feed their own sources
  • A pilot with a real broadcaster, independently measured
Why it matters

For journalism, and for the public square.

Small newsrooms cannot afford a full team — a researcher who retrieves a fact in seconds, a fact-checker, a devil's advocate, an analyst. A deliberative panel delivers these as a tool. It does not replace the journalist; it amplifies them. The live host stays master of the broadcast — able to interrupt, redirect, overrule.

And it serves the wider information environment three ways: fact-checking caught at the moment of utterance, when disinformation still outruns its correction; reasoning made transparent, so the format teaches the media literacy it practises; and structured disagreement as an antidote to AI content that manufactures consensus — a format built on mandatory collision cannot be turned into a machine for engineering agreement.

Responsible by design

AI on air is a risk. The format is built around it.

AI enters a live transmission before a mass audience only under control: the human host can interrupt and silence it at any moment; the system is biased toward silence when uncertain, rather than filling air with confident error; and the AI never impersonates a human — it is always identified as AI.

This is not a feature bolted on at the end — it is the core. The aim is not to add one more uncontrolled source of synthetic content to the environment, but to demonstrate a responsible model for the presence of AI on air. How to bring AI into live broadcast without eroding trust is, in itself, a question of public-interest research.

Where it stands

A proven format, seeking the support to become a measured public good.

The working platform has demonstrated its core — debates under a live or AI host, one to seven experts, addressed follow-ups, and the hearing of human speech, in two languages. The next step is a pilot with a real broadcaster and an independent evaluation: whether audiences notice and value real-time fact-checking, and whether deliberative segments hold attention differently from conventional ones.

Led by a media team with three decades in international public broadcasting.