The year of generative AI 😱

Will AI systems change everything, or end up being nothing more than a bit of a content generating assistance? 🧐 We cannot be sure yet. But 2023 has been the year of generative AI, the year in which technology to process language and images fully arrived, and many, many humans, me among them, tried to understand how they work and what their capabilities – the fact that they can calculate meaning – might actually mean. These two texts are among those that explore this.

I have been fascinated by the weird way in which AI systems approach our world and generate their intelligence early on, and this is what the first text is about. Errors in machine learning systems are an interesting phenomenon. Obscuring the role of AI as a shiny tool of capital, the phenomenon is usually studied with the aim to overcome those errors. My first text, Errors are no Exception: On the Alien Intelligence of Machine Learning, aims for another direction by digging into one particular error, the phenomenon of “adversarials”.

The second text, Thinking Through Generated Writing, is the outcome of my sabbatical, which I spent with understanding how Large Language Models work and what this might mean for writing. For this, the I returned to the first release of ChatGPT to study the mechanics of Large Language Models (LLMs) and to counterpose them against established theories of language and writing. My aim was to explore the hypothesis that ‘generated writing’ is “the beginning of new writing”, a writing that has its own cultural logic and tendencies. To show this, I revisited philosophical theories of writing, in particular those of Jacques Derrida and André Leroi-Gourhan, and reads them in light of the computational methods of LLMs to understand in what way ‘generated writing’ diverges from the writing we have known before.

I have to say, writing about them and with them since a year was good. I look forward to more.
Happy New Year to AI & us! 🥂