Editorial AI AI Oh
By Toh Hsien Min
A bit of a pass-the-parcel developed during the production of this issue, specifically on the editorial. "Anyone have a view on the NLB AI kerfuffle?" I asked my co-editors. "If any of you have anything to say... wanna do a "pop-up" or "takeover" of the editorial?" It didn't take long for the parcel to come back. "Don't think there's more to say at this point regarding AI," one said. "I'm unable," said another, "as I'm preparing for a work trip. Why don't you write about bidding for art instead?" Reluctantly I pulled up the articles on members of Singapore's literary community issuing a statement criticising the National Library Board (NLB) for its stance on GenAI, and indeed the statement itself. Reading about the concerns on NLB's "uncritical endorsement of generative AI", I couldn't help but wonder whether it was ChatGPT or Claude that was used in writing the statement. (It couldn't have been DeepSeek, which burst into the popular consciousness only a couple of weeks after the 7 Jan statement.) Doesn't "We urge policymakers to exercise greater caution in the adoption of this technology, at the risk of permanently damaging Singapore's literary landscape" sound written by a machine? To be fair, I don't suppose it to be wrong to be concerned about intellectual property or the environment. It's just that the arguments don't really stack up when you take a close look at them. The NLB have come out explicitly to say its StoryGen prototype uses "materials... within NLB's copyright, appropriately licensed, or no longer under copyright"; but the statement says, effectively, we don't care, we hold you responsible for the sins of others. The statement continues by moaning that ""AI" literature trends towards mediocrity, without originality". But isn't that a good thing for flesh-and-blood writers? If GenAI were to become capable of writing as well as Alice Oswald or Ian McEwan, would the argument flip towards wholesale adoption of AI? One suspects not. And as I've already written elsewhere, it seems that reading other people's writing in order to come up with one's own writing is okay when it's done by a human but not by a machine. I do not wait with bated breath to see how the anxiety of influence will be cited to justify the former. Rather, I think about my first encounters with what was then called "strong AI" during the early 1990s, which, unless I misremember, was prompted by none other than Prof Edwin Thumboo, who circulated photocopied articles on the debates between the likes of Hans Moravec and Roger Penrose. These touched in part on the proposition that humanity might eventually achieve a form of immortality as a robot brain surgeon gradually replaces the failing parts of your brain by transferring their activity to a mechanical equivalent. At some point the whole brain might be replaced, and if you continue to otherwise behave as you did, who's to say where that dividing line between a human and a machine lies? That transmigration for me is where the horror is. A writing machine that can churn out pastiche? Aside from being an instance of Moravec's broadly accurate prediction that human-equivalent personal machines would be in place by 2030, that's the small stuff. Don't sweat the small stuff. Having written all that, I see I've missed a trick. I should've passed this parcel to GenAI instead. Part of me wants to comment on the humanity of the writing in this issue (leaving aside the one poem supposedly written by bread), but part of me doesn't think I should. The reason for the latter is that it feels as meaningful as to talk about the sweetness of sugar. Instead, I would rather quote the killer first line from Tammy Ho's poem, 'Close Enough': "It is almost always when I am home alone that the people upstairs move their heavy furniture." (That or maybe bringing the Althusserian interpellation highlighted in Ow Yeong Wai Kit's essay to bear on the debate at hand.) Ultimately it's how we respond to situations before us that gives us our humanity. The rest of it is machinery. QLRS Vol. 24 No. 1 Jan 2025_____
|
|
|||||||||||||
Copyright © 2001-2025 The Authors
Privacy Policy | Terms of Use |
E-mail