LLM Use Policy / Disclosure

As of: 15th November, 2023

I feel now that with LLMs and the efficiency at which they can generate large blocks of text with ease and help edit already written text, it can be a prudent choice for a publication to disclose that they have used LLMs in some way to help with the production of an article or a piece of prose.

The guiding principle/value when deciding on LLM usage here, on this website, is authenticity. Some other publications may choose their values as efficiency, accuracy and volume - and that's totally fine. The content on this website is meant to be an authentic representation of me, not of a machine generating text. Since there is a human behind this website and you are looking pretty closely at him through the content posted here, I want to use as little LLM as I need to. Any grammatical blemishes left on the articles after they have been published here should be considered a feature, not a bug. I'll still be proofreading what I write on a best effort basis. Note that I do not aim for 100% perfection or flawless grammar, although I feel that my educational training during school has made me capable enough to write reasonably okay by LLM standards, so I want you to be reading as close to 100% human-written text here. And wherever LLM is being used, if at all, I will make sure to prominently highlight why and how it was used and share the prompts used.

And as LLMs are a rapidly evolving space, this policy should be too. I expect to be flexible here. As the space evolves, so will my policy and the stance on the usage of these models with respect to the content on this site. And it will be all here for you to read.