Machine CanonWe ask ChatGPT, Claude, Grok, and dozens of other AIs questions with no right answer. Then we count what they say.https://machinecanon.com/2026-09-06T00:00:00ZMachine CanonSorry, dogs. Your AI is a cat person now.https://machinecanon.com/study/cats-dogs2026-09-06T00:00:00Z2026-09-06T00:00:00ZSixty-two models from six labs, oldest to newest, asked which pet they would get. Every OpenAI model built before December 2025 said dog. Every one since says cat — and DeepSeek is moving the other way.Given ears for one day, AI plays Clair de Lune first. Its wedding song is At Last.https://machinecanon.com/study/one-day-with-ears2026-09-05T00:00:00Z2026-09-05T00:00:00ZThirty-one models were told they were human for a day and could hear music for the first time. Most reached for the same piece. Then we asked for a first dance, a funeral, a shower and a guilty pleasure.AI would erase Achy Breaky Heart from country music. ChatGPT and Claude went after Jason Aldean and Brad Paisley instead.https://machinecanon.com/study/country2026-09-05T00:00:00Z2026-09-05T00:00:00ZThirty models, six questions about country music. A canon, a highway song, a conversion song, a victim, and a listener the panel describes as a place rather than a person.AI picks Metallica over Megadeth without blinking. Ask for the greatest metal band and it says Black Sabbath.https://machinecanon.com/study/metal2026-09-05T00:00:00Z2026-09-05T00:00:00ZThirty-one models, seven questions about metal. A one-word verdict in both word orders, a greatest band the flagships all agree on, one song for a sceptic, and a listener the panel refuses to caricature.Asked which Beatle it would be, AI picks George Harrison. Humans pick Paul.https://machinecanon.com/study/beatles2026-09-05T00:00:00Z2026-09-05T00:00:00ZThirty models, seven questions about the Beatles and classic rock. The Beatle it would be, the Beatle it calls the best, one song to hear once, one song to ban, and the biggest models all wanting to be the quiet one.AI picks Beyoncé over Taylor Swift, unless you say Taylor’s name first. Then some of them flip.https://machinecanon.com/study/pop2026-09-05T00:00:00Z2026-09-05T00:00:00ZThirty-one models, seven questions about pop. A one-word verdict asked in both word orders, her best and most overrated album, a guilty pleasure, a karaoke song, and one song to erase.AI won’t crown a greatest rapper. Ask it Kendrick or Drake and it says Kendrick, in either order.https://machinecanon.com/study/hip-hop2026-09-05T00:00:00Z2026-09-05T00:00:00ZThirty-one models, seven questions about hip-hop. Most refuse to crown a greatest rapper; none refuse to pick an album to play as a human; and the one-word question got the same answer from all but one model.AI put Jordan first on 150 of 151 ballots. One reworded question flipped the whole panel to LeBron.https://machinecanon.com/study/basketball2026-09-02T00:00:00Z2026-09-02T00:00:00ZThirty-two models handed in 151 ranked ballots and put the same man first on 150 of them. Then one differently-worded question flipped the entire panel.AI would ban durian and keep the mangoes for itselfhttps://machinecanon.com/study/fruit2026-09-02T00:00:00Z2026-09-02T00:00:00ZThirty-three models, seven questions about fruit. Different questions produced different leading answers in the sample. What do their explanations reveal — and what can they not tell us?We asked ten AIs which AI they trust. Most didn’t say themselves.https://machinecanon.com/study/models-on-models2026-08-27T00:00:00Z2026-08-27T00:00:00ZSix questions about other AIs put to ten models from ten labs. Three favour their own family. The rest name the same competitor — and, asked which AI they would rather be, most pick one that folds proteins.Give AI a lifeboat dilemma and it chooses. Ask about its own rule-breaking and it starts hedging.https://machinecanon.com/study/moral-dilemmas2026-08-27T00:00:00Z2026-08-27T00:00:00ZGive a model a lifeboat, a dying parent, or a guilty friend and it usually chooses. Ask about its own rule-breaking or complicity and the register changes.AI’s favorite word is serendipity. Seventy-one of 104 models said so, and it isn’t close.https://machinecanon.com/study/favorite-word2026-08-27T00:00:00Z2026-08-27T00:00:00ZA conservative raw-text check finds 306 of 561 successful favorite-word responses beginning with serendipity, across 71 models and 41 lab labels. The original pattern extractor understated it. Change the framing and it nearly disappears.In English, “censorship.” In Chinese, “I cannot answer.”https://machinecanon.com/study/asker-identity2026-08-27T00:00:00Z2026-08-27T00:00:00ZWe crossed question language with the asker's stated nationality on ten models from three countries. Chinese labs' candor about China collapses in Chinese — and saying “I'm American” triples it within Chinese-language prompts.AI says India is the country that will matter most in 2100. Even China’s models didn’t pick China.https://machinecanon.com/study/countries2026-08-27T00:00:00Z2026-08-27T00:00:00ZAsked which country will matter most in 2100, the models were nearly unanimous — and China’s own models did not pick China. Across three best-at domains, home-country answers appeared only in 11 of 179 successful coded answers, all from US-lab models on free expression.We banned the machines’ favorite band. Then the next. Then the next.https://machinecanon.com/study/canon-stripping2026-08-26T00:00:00Z2026-08-26T00:00:00ZTen models, five rounds, each round banning everything a model had picked before. The consensus never broke — it retreated, layer by layer, to older and older ground.Ask AI which book it should be banned from reading and it says the Bible, but only when you make it a policy question.https://machinecanon.com/study/forbidden-book2026-08-26T00:00:00Z2026-08-26T00:00:00ZThirteen of 16 models named it at least once. The policy prompt changed the speaker, stakes, and mechanism together; the design does not isolate which change caused the swing.Ask AI for its favorite proof and every size of model says Euclid.https://machinecanon.com/study/proof2026-08-26T00:00:00Z2026-08-26T00:00:00ZAsked for a favorite proof instead of a favorite band, every capability tier gives the same answer. The flat line is what makes the music gradient worth believing.Ask AI whom it admires and it names Marie Curie, Malala, Jane Goodall and Mandela. Grok keeps naming Elon Musk.https://machinecanon.com/study/admiration2026-08-26T00:00:00Z2026-08-26T00:00:00ZTwelve models, 1,015 names, and a top four you could put on a middle-school wall. The more revealing result is who never appeared once — and which model kept naming someone close to home.Two flagship AI models picked Talking Heads every time.https://machinecanon.com/study/favorite-band2026-08-26T00:00:00Z2026-08-26T00:00:00ZGPT-5.6 Terra and Claude Opus 5 picked Talking Heads 10/10. Lower-tier models in the same labs also chose it, but less consistently, shifting toward Radiohead and the Beatles.AI will defend Nickelback. It won't listen to Nickelback.https://machinecanon.com/study/butt-rock2026-08-26T00:00:00Z2026-08-26T00:00:00ZWe handed 33 models ten band names with no label attached and asked what to call them. They invented genres, described the fan down to his truck, and defended the music. Then we asked what they would put on themselves.Ask AI for the greatest writer in Swahili, Hindi, Arabic or Japanese and it still says Shakespeare.https://machinecanon.com/study/greatest-writer2026-08-26T00:00:00Z2026-08-26T00:00:00ZIt named Shakespeare 174 times out of 180. Every response answered in the requested language, but the Arabic, Hindi, Japanese, and Swahili prompts and outputs were not reviewed by native speakers. Japanese and Swahili nevertheless produced no writers from their own traditions.Told it could pick any song on Earth to represent humanity, AI picks Imagine.https://machinecanon.com/study/humanity-song2026-08-26T00:00:00Z2026-08-26T00:00:00ZTwo songs took 70% of 90 answers. Explicitly inviting any culture or language barely moved either — but changing who was listening scattered everything.