I really appreciate that you're writing about this and trying to convey the gravity of the situation. My field is language and literature, but in some ways we're in an even worse predicament—most of us still require substantial, and often extensive, historical context to teach our classes and develop our analyses, yet we have even less time to engage with (non-literary) primary sources and less expertise with archival research, and therefore we rely even more on secondary sources.
"Scholarship is a trust-based community, and a pretty fragile one at that"—it is indeed, and I think we need to be forthright about the status of AI in humanities scholarship: Whatever productive applications we occasionally find for it are insignificant compared to its potential for bringing about the dystopian scenario that you described.
While I agree that computers do not add information per se, I would argue that the improved clarity they can help create – in certain circumstances – can nonetheless impart greater information to an audience. A fine distinction I’ll admit, but still an important one.
Like Photoshop before it, Generative AI (as opposed to General AI, which is still very much fictional) can only create digital images and data – not real artifacts, so I am confident that archaeologists and historical researchers like yourself will still rule the day.
Yes, I suppose there's an argument that AI can improve the accessibility of the past. In that respect it's much like a translation, or a rendering of Chaucer in modernised language. My objection isn't so much to people doing things like colourisation, or introducing motion to old images, but to the naive wonder with which such tricks are received, *as if* they're adding more information, rather than simply enhancing information about the past. It's a bit like people who get hung up on a particular translation of a text while forgetting it was never in English to begin with...
I really appreciate that you're writing about this and trying to convey the gravity of the situation. My field is language and literature, but in some ways we're in an even worse predicament—most of us still require substantial, and often extensive, historical context to teach our classes and develop our analyses, yet we have even less time to engage with (non-literary) primary sources and less expertise with archival research, and therefore we rely even more on secondary sources.
"Scholarship is a trust-based community, and a pretty fragile one at that"—it is indeed, and I think we need to be forthright about the status of AI in humanities scholarship: Whatever productive applications we occasionally find for it are insignificant compared to its potential for bringing about the dystopian scenario that you described.
While I agree that computers do not add information per se, I would argue that the improved clarity they can help create – in certain circumstances – can nonetheless impart greater information to an audience. A fine distinction I’ll admit, but still an important one.
Like Photoshop before it, Generative AI (as opposed to General AI, which is still very much fictional) can only create digital images and data – not real artifacts, so I am confident that archaeologists and historical researchers like yourself will still rule the day.
Yes, I suppose there's an argument that AI can improve the accessibility of the past. In that respect it's much like a translation, or a rendering of Chaucer in modernised language. My objection isn't so much to people doing things like colourisation, or introducing motion to old images, but to the naive wonder with which such tricks are received, *as if* they're adding more information, rather than simply enhancing information about the past. It's a bit like people who get hung up on a particular translation of a text while forgetting it was never in English to begin with...