I just received a printed copy of the latest ACM Interactions magazine, where my article is featured as the cover story: Reclaiming the Computer through LLM-Mediated Computing.

I have been thinking about writing something for Interactions about LLMs for quite some time. They are most commonly described as examples of generative AI, but to me that sells them short. LLMs are not just algorithms that can produce artifacts that previously required people. They do much more. Perhaps most importantly, they can process inputs, infer intent within context, and act on that. Calling them “generative” emphasizes the output, whereas I am more impressed with how they work with the input. To me they offer to serve as a very interesting material for interaction design.

In this article, I outline the idea of LLM-mediated computing: a mode of computing where LLMs infer human intent and generate code in response, making the computer’s capabilities emerge through interaction rather than being predefined in applications. This reframes the computer not as a static collection of tools, but as something that dynamically discloses its possibilities through ongoing interaction between the user, the LLM, and the machine.

I am very happy to see this article featured as the cover story in Interactions and grateful to the editors for granting this honor.