The historian Francis Young recently published a very interesting essay on the threat that generative AI poses to certainty in historical scholarship and, particularly, to the ability of historians to trust the secondary sources on which they must often rely. If, as is evidently already the case, AI-generated text is poisoning the well of historical scholarship, how can we know where to place our trust? As Young puts it, “In an ideal world we would all check and double check everything against the archive, of course; but that isn’t the reality. Things need to get published, discussion needs to continue; and we are all reliant on the purity of the chain of information supply to allow us to do our job”.

In the course of this long, long project I have perhaps devoted more time than is normally possible to the task of checking and double checking my sources. Partly this is because I feel a sense of obligation that what I have to say about Macintosh and his world is as well supported as it possibly can be, but it is also a function of encountering what are, for me, new areas of historical scholarship, each with their own certainties already cemented in the secondary literature. I am, in general, interested in how certain ideas become fixed in the historiography and have found, on many occasions, that unverified or inaccurate claims have circulated in the literature apparently effortlessly, becoming established through repetition as historical fact. Sometimes these can arise from something as minor as a transcription error, or the conflation of two or more individuals with similar names.

Tracing something back to the archive is not, as we know, a guarantee of certainty; primary sources are not, in and of themselves, sources of “truth”, but they are part of the foundation on which a certain type of rigour in scholarship depends. Rather than seeking to separate right from wrong in the process of checking and double checking, I find myself more interested in why it is that certain interpretations emerge, why certain stories become canonical, and what effect those canonical analyses have on the scholarship that comes afterwards. All that being said, there is an undeniable satisfaction in being able to know something with certainty (or sufficient certainty), and that most often comes from following the “chain of information” to its original source.

This was the case recently when I was able to identify, for the first time, the date on which Macintosh’s onetime ally, and former Governor of Madras, John Whitehill had died. The latter part of Whitehill’s life, spent mostly in France and significantly disrupted by the Revolution, has long been enigmatic. Driven by nothing more noble than bloody-mindedness, I have tried at intervals during the last couple of years to find some definitive evidence of Whitehill’s date of death. He was not, like Macintosh, an obscure figure, and was well known in India, Britain, and France, and I always found it puzzling that a basic piece of biographical information was so elusive.

This week I located a reference to Whitehill in the records of the Parisian notary Alexandre Pierre François Robert-Dumesnil, showing that Robert-Dumesnil had prepared an inventory, on 16 June 1812, of Whitehill’s possessions following the latter’s death. While the record did not include the date of Whitehill’s death, it did provide his address: 126 rue de Vaugirard.

With an address in hand it was possible, following the helpful advice of the UCL historian Simon Macdonald (who has separately identified Whitehill’s date of birth), to find the corresponding record of Whitehill’s demise in the table des décès for the old 10th arrondissement.

Here we can see that Whitehill died on 15 April 1812, his profession listed as rentier (one who lived off the income of his capital investments).

While it would be easy to say that none of this really matters, I think that the new disinformation age ushered in by generative AI means that these things really do matter in insuring the “purity of the chain of information supply” that Young invokes. In that spirit, I have updated John Whitehill’s Wikipedia entry with his date of death, one link in the chain secured.