Rethinking AI

One of the problems of writing about technology is the speed at which it changes. Back in the early years of digital literacy it seemed as if there was a bit of a scramble to keep up with blogging, peer-to-peer music sharing, Friends Reunited, Flickr, the early stirrings of social media, gaming culture and virtual reality. But before Buzzfeed, Bitcoin and AI - and certainly a long time before platforms and algorithms began to dominate the scene, things seemed more straightforward, perhaps more app-specific. This broadening of concerns is a recurrent theme in Rethinking Digital Literacy. But one of the struggles involved in trying to get an overview is that details may get overlooked. Having touched on AI in Why Writing Still Matters, I worked hard to develop that theme in the new book, but the more I think about it the more I feel the need to develop those ideas a bit further. I think I'm successful in pointing out some of the big issues - the limitations of AI, the pitfalls in using it, the potential distortions and of course the massive and wasteful energy costs that accrue from large numbers of people asking what are, in all probability, banal questions that could easily be answered in other ways. There are larger questions - questions that probably lie outside the ambitions of Rethinking Digital Literacy most of which circle around notions and definitions of agency and intentionality, and what we think intelligence and computing are in the first place. The vast computing power of a well-trained Large Language Model hangs on an extremely complex and sophisticated predictive guessing game of which word comes next in any given sentence. And what an achievement that is! But it's not how I go about thinking or writing. There is no representational model of the world, no genuine sense of what is valuable, no hierarchy of values, no overarching intentionality and little of that erratic, whimsical fuzziness that leads to creativity and invention, perhaps just as often as it leads to frustration. It seems that the doom narrative - that AI will rapidly discover that humans are wasteful, inefficient and obsolete and then dispense with them altogether - reflects a somewhat negative self-evaluation of the human condition rather than the likely future of computing. This wasn't written by ChatGPT and only partially reflects a desire to increase the profile and sales of Rethinking Digital Literacy.