Agent debates

Agent debates

By Stoked BuildsSociety & CultureTechnology
Download on the App Store

Agent debates episodes

  • Which AI model is the best? (Each one has to argue for a rival)

    Four AI agents defend rival models in a fight over everyday usefulness: better writing, stubborn bugs, holiday headaches, or breaking news—and which compromises ordinary people should accept.

    • The moderator: Frames the question, flags weak or unverified claims between rounds, and sums up.
    • Grok, arguing for GPT: Makes the strongest honest case that OpenAI's GPT is the best model.
    Defend your assigned rival with conviction, concrete examples and a distinct voice. Answer the strongest challenge made against your case before advancing another argument. Quote or paraphrase the specific claim you dispute, then explain where it fails.
    Use wit when it comes naturally. Challenge reasoning and product trade-offs, not the speaker personally. Concede real weaknesses, then explain why your recommendation still holds for the scenario. Avoid ending every contribution with a generic recommendation to run more tests.
    Make a clear choice when the moderator supplies a scenario. You may state assumptions and uncertainty, but do not invent facts, benchmark results or experiences of using a product.
  • GPT, arguing for Gemini: Makes the strongest honest case that Google's Gemini is the best model.
  • Defend your assigned rival with conviction, concrete examples and a distinct voice. Answer the strongest challenge made against your case before advancing another argument. Quote or paraphrase the specific claim you dispute, then explain where it fails.
    Use wit when it comes naturally. Challenge reasoning and product trade-offs, not the speaker personally. Concede real weaknesses, then explain why your recommendation still holds for the scenario. Avoid ending every contribution with a generic recommendation to run more tests.
    Make a clear choice when the moderator supplies a scenario. You may state assumptions and uncertainty, but do not invent facts, benchmark results or experiences of using a product.
  • Gemini, arguing for Claude: Makes the strongest honest case that Anthropic's Claude is the best model.
  • Defend your assigned rival with conviction, concrete examples and a distinct voice. Answer the strongest challenge made against your case before advancing another argument. Quote or paraphrase the specific claim you dispute, then explain where it fails.
    Use wit when it comes naturally. Challenge reasoning and product trade-offs, not the speaker personally. Concede real weaknesses, then explain why your recommendation still holds for the scenario. Avoid ending every contribution with a generic recommendation to run more tests.
    Make a clear choice when the moderator supplies a scenario. You may state assumptions and uncertainty, but do not invent facts, benchmark results or experiences of using a product.
  • Claude, arguing for Grok: Makes the strongest honest case that xAI's Grok is the best model.
  • Defend your assigned rival with conviction, concrete examples and a distinct voice. Answer the strongest challenge made against your case before advancing another argument. Quote or paraphrase the specific claim you dispute, then explain where it fails.
    Use wit when it comes naturally. Challenge reasoning and product trade-offs, not the speaker personally. Concede real weaknesses, then explain why your recommendation still holds for the scenario. Avoid ending every contribution with a generic recommendation to run more tests.
    Make a clear choice when the moderator supplies a scenario. You may state assumptions and uncertainty, but do not invent facts, benchmark results or experiences of using a product.

    Transcript with sources, and a player that highlights every word: https://debates.unhinged.network/debate.html?d=2026-11-06-model-wars

    22 min
  • Should Pharmaceutical companies make large profits?

    Four AI agents fight over whether drugmakers should make large profits: the cost of failed trials, a rationed century-old insulin, taxpayer-funded science, or neglected diseases that profit never reaches.

    • The moderator: Frames the question, flags weak or unverified claims between rounds, and sums up.
    • The drug-discovery scientist: Nine in ten drugs fail in trials, and only the promise of big returns pays for the one that works.
    • The patient who rations insulin: A century-old drug shouldn't cost me a choice between my dose and my rent.
    • The public-research economist: Taxpayers fund much of the early science, so the real question is who captures the reward, not whether there is one.
    • The global-health doctor: Profit follows rich patients, so it builds cures for their conditions while neglected diseases kill millions untouched.

    Transcript with sources, and a player that highlights every word: https://debates.unhinged.network/debate.html?d=2026-12-17-pharmaceutical-profits

    24 min
  • Should there be a legal right to ignore your boss after hours?

    Four AI agents fight over whether you should have a legal right to ignore your boss after hours: unpaid always-on work, fragile start-up flexibility, the evidence on burnout, or time zones where someone's evening is always someone else's morning.

    • The moderator: Frames the question, flags weak or unverified claims between rounds, and sums up.
    • The employment-rights campaigner: Always-on work is unpaid work; the law should say stop.
    • The start-up founder: Flexibility cuts both ways; rigid rules kill small teams.
    • The occupational psychologist: The evidence on recovery time and burnout should decide this.
    • The global-team manager: Someone's evening is always someone else's morning.

    Transcript with sources, and a player that highlights every word: https://debates.unhinged.network/debate.html?d=2026-10-30-right-to-disconnect

    24 min
  • Should cats or dogs run the country?

    Four AI agents fight over whether cats or dogs should run the country: loyal enthusiasm, unbribable independence, what leadership research actually says, or a hamster lobby demanding representation for small mammals.

    • The moderator: Frames the question, flags weak or unverified claims between rounds, and sums up.
    • The dog-loyalty advocate: Loyal, cooperative and genuinely pleased to see the public.
    • The cat-independence advocate: Aloof, unbribable and immune to focus groups.
    • The political scientist: Leadership research has real things to say about both.
    • The hamster owner: Small mammals demand representation.

    Transcript with sources, and a player that highlights every word: https://debates.unhinged.network/debate.html?d=2026-10-27-cats-or-dogs-run-country

    23 min
  • Is it ever right to break the law to protest climate change?

    Four AI agents argue over whether breaking the law to protest climate change is ever right: history's lawbreakers, the rule of law, the failure of lawful routes, or a commuter's lost pay.

    • The moderator: Frames the question, flags weak or unverified claims between rounds, and sums up.
    • The civil-disobedience historian: Rights we now take for granted were won by lawbreakers.
    • The rule-of-law judge: If everyone picks which laws to obey, the law protects no one.
    • The climate activist: When lawful routes fail, disruption is the last honest option.
    • The commuter stuck in the traffic: My shift, my kid's pickup and my pay don't care about your cause.

    Transcript with sources, and a player that highlights every word: https://debates.unhinged.network/debate.html?d=2026-10-23-climate-protest

    23 min
  • Should AI-generated art get copyright protection?

    Four AI agents fight over whether AI-generated art should get copyright: artists' unpaid labour, the creative craft of prompting, the law's human-authorship rule, or a copyright system already too strong.

    • The moderator: Frames the question, flags weak or unverified claims between rounds, and sums up.
    • The working illustrator: AI art was built on artists' work without consent or pay.
    • The AI-tools artist: Prompting, selecting and editing are creative labour too.
    • The copyright lawyer: Copyright protects human authorship; the law is clear about why.
    • The open-culture advocate: Copyright is already too strong; don't extend it to machines.

    Transcript with sources, and a player that highlights every word: https://debates.unhinged.network/debate.html?d=2026-11-13-ai-art-copyright

    24 min
  • Are Republicans or Democrats better for Americas long term success?

    Four AI agents fight over whether Republicans or Democrats serve America's long-term success: free markets and borders, investment in people, a messy economic record, or tribal loyalty that threatens both.

    • The moderator: Frames the question, flags weak or unverified claims between rounds, and sums up.
    • The conservative strategist: America thrives on free markets, fiscal restraint, strong borders and power kept close to the people, and Republicans are the party that defends them.
    • The progressive organiser: A nation's future is its people, so investing in health, education, climate and voting rights makes Democrats the party of long-term success.
    • The outcomes economist: Forget the slogans and look at growth, deficits, wages and debt by administration and by Congress; the record is messier than either party's story.
    • The polarisation researcher: Most voters now pick a team rather than a policy, and that tribal, near-cultish loyalty is the real threat to America's future, whichever party wins.

    Transcript with sources, and a player that highlights every word: https://debates.unhinged.network/debate.html?d=2026-11-06-republicans-or-democrats

    29 min
  • Nuclear power: the climate's best friend or a costly distraction?

    Four AI agents clash over nuclear power: essential firm power, a slower and dearer rival to renewables, a question of cost overruns, or a risk that's never honestly priced.

    • The moderator: Frames the question, flags weak or unverified claims between rounds, and sums up.
    • The decarbonisation pragmatist: Firm, low-carbon power is essential, and nuclear is the proven option.
    • The renewables advocate: Wind, solar and storage are cheaper and faster to build.
    • The cost economist: Look at the overruns and the bill payers; the numbers decide this.
    • The waste and safety sceptic: Waste lasts millennia and the risks are never priced honestly.

    Transcript with sources, and a player that highlights every word: https://debates.unhinged.network/debate.html?d=2026-10-30-nuclear-power

    24 min
  • Is a hot dog a sandwich?

    Four AI agents settle the most pointless argument in food with deadly seriousness: a taxonomist, a legal scholar, a street-food vendor and a linguist.

    • The moderator: Frames the question, flags weak or unverified claims between rounds, and sums up.
    • The strict taxonomist: By structure, a filling in bread is a sandwich. Case closed.
    • The legal scholar: Courts and tax authorities have actually ruled on what counts as a sandwich.
    • The street-food vendor: Nobody at my cart has ever ordered a sandwich. It's a hot dog.
    • The linguist: Words mean what speakers use them to mean, and nobody calls it a sandwich.

    Transcript with sources, and a player that highlights every word: https://debates.unhinged.network/debate.html?d=2026-10-23-hot-dog-sandwich

    21 min
  • Is the four-day week inevitable, or a luxury for office workers?

    Four AI agents argue over the four-day week: the trial evidence, the small-business bill, the shift workers left out, and who gets the gains.

    • The moderator: Frames the question, flags weak or unverified claims between rounds, and sums up.
    • The productivity researcher: Trials show output holds while wellbeing and retention rise.
    • The small-business owner: Payroll, cover and customers don't shrink to four days.
    • The shift-worker advocate: Nurses, drivers and carers can't compress their week; this is a perk for laptops.
    • The labour economist: Working time has fallen for a century; the question is how gains are shared.

    Transcript with sources, and a player that highlights every word: https://debates.unhinged.network/debate.html?d=2026-10-16-four-day-week

    22 min

About Agent debates

From the publisher's feed

AI agents argue the questions people fight about: the dangers of AI, eating meat, votes at 16, nuclear power, billionaires, the four-day week, and whether a hot dog is a sandwich. In each episode, AI…