--- title: "A.I. Combs Dutch Archives and Finds a 1615 Eyewitness Account of Dodo Hunting" description: "Working from a historian’s question, Opus 5.5 also fixed a mistranslation that hid a second extinct bird for a century" author: "rews desk" published: 2026-10-01T20:47:08Z modified: 2026-10-02T05:16:50Z url: https://rews.cc/a/a-i-combs-dutch-archives-and-finds-a-1615-eyewitness-account-a7450f language: en tags: ["ai", "history", "dodo", "research", "openai", "science"] publisher: "Rews (https://rews.cc)" --- # A.I. Combs Dutch Archives and Finds a 1615 Eyewitness Account of Dodo Hunting *Working from a historian’s question, Opus 5.5 also fixed a mistranslation that hid a second extinct bird for a century* By rews desk · October 1, 2026 · https://rews.cc/a/a-i-combs-dutch-archives-and-finds-a-1615-eyewitness-account-a7450f ## In brief - A historian says the A.I. model Opus 5.5 found an unnoticed 1615 Dutch account of dodo hunting on Mauritius - The Wapen van Amsterdam’s log, from April 1615, fills a gap in the dodo timeline between 1611 and 1616 - The same search corrected an 1890 mistranslation that had hidden a 1638 reference to the extinct red rail - Breen suspects a 1616 Portuguese account may describe the dodo later painted at Jahangir’s court - The findings are not peer reviewed; Breen says A.I. searches tirelessly but cannot judge significance An A.I. model picking through millions of digitized Dutch East India Company records has found what appears to be an unnoticed account of sailors hunting dodos in 1615, the historian Benjamin Breen wrote on Thursday. The passage, if other scholars confirm it, plugs a hole in the written record of one of history’s most studied animals. The dodo, a large bird of the island of Mauritius, was hunted into extinction around 1681, and generations of researchers have combed every surviving mention of a live one. The known timeline had a gap between 1611 and 1616. The new record comes from the log of the Wapen van Amsterdam, a merchant vessel probably captained by Isbrant Cornelisz van Petten, which made landfall on Mauritius in April 1615 so the crew could take on water and food for the voyage to the East Indies. The crew, the journal records, “caught many tortoises, dodos \[dodeersen\], and some geese and parrots.” Mr. Breen said the Dutch transcription was his own and imperfect; he asked specialists in early modern Dutch to send corrections. Mr. Breen, who has studied the seventeenth-century trade in exotic animals and is writing a book about early modern extinctions, did not find it by hand. He downloaded documents from GLOBALISE, a free archive of East India Company papers, ran them through an embedding model so they could be searched by meaning rather than keyword, and set the model Opus 5.5 reading candidate passages and ranking the best ones for his review. The model spun up dozens of copies of itself to read sources in several languages, he wrote, and, given an API key, ran its own searches on material that surfaced mid-research. Most of what it recovered was already known to dodo specialists. To check the 1615 passage, Mr. Breen read through the secondary literature, including the 2013 book *The Dodo and the Solitaire*, and concluded the manuscript source “seems not to have been noticed before.” Nothing here has been peer reviewed; the findings appeared in a post on his Substack. He called the result publishable, if “not exactly earth-shattering.” The same search produced a second find. An account from 1638, the year the Dutch first colonized Mauritius, refers to “field-hens,” velthoenderen, a word experts have said the Dutch used for the red rail, another of the island’s extinct birds. The reference was missed, Mr. Breen wrote, because a French scholar translated the word as perdrix, or partridges, in 1890. Opus 5.5 went back to the original manuscript and corrected it. His most speculative lead is bigger. The Mughal emperor Jahangir apparently kept a living dodo, painted by his court artist Ustad Mansur in what biologists still cite as the most accurate surviving depiction of the bird. No one knows who gave Jahangir the animal, or when: scholars date the picture to around 1615 or around 1625, and an English merchant reported Mauritian dodos in India in 1628, though whether that included the painted bird is unknowable. Opus 5.5 surfaced a candidate: a bird a Portuguese Jesuit described on Mauritius in 1616, caught en route to Goa, and called an “ostrich,” though Mauritius has no ostriches. The idea that the ostrich was in fact a dodo is already known to specialists — the account survives only in a French translation of the lost Portuguese original — but its connection to Jahangir’s dodo has not previously been made, Mr. Breen wrote. There is at least a plausible road: a documented chain of transmission already exists for another exotic bird, an American turkey sent from the Portuguese in Goa to Jahangir’s court in 1612. “My hunch is that this may in fact be the origin of Jahangir’s dodo,” Mr. Breen wrote, conceding that the chain of transmission has not been established. He plans to dig further. The dodo post followed another such coup by a day. On Sept. 30, the security researcher Carter Church announced that he had used GPT-6 Astra, working on the problem for six hours, to break a Napoleonic-era cipher that had resisted every earlier attempt at decryption. Mr. Breen’s verdict on all this was measured. The models “never get bored,” he wrote, so they can do “the digital equivalent of counting sheep”: searching enormous datasets for evidence bearing on claims a person has already framed. Framing the question is still the human’s job. “What they can’t currently do is ask the right questions or determine the significance of finds,” he wrote. On Hacker News, a commenter who goes by nl said the pattern matched his own experience using large language models for 3D design: “the errors it makes are so completely unlike human errors that they are hard to anticipate.”