At a company where most communication is written, your writing is part of how people know you. In my work about 90% of the communication is written via Slack, P2 (internal blogging system), Github and more.
Just like when you read a book from your favourite author, or a blog post from your favourite blogger, you come to know a bit about them and their personality. How they write, how they think, what their sense of humour is like. Writing is a creative outlet after all, and our personality is often expressed through these creative outlets.
I am sure you can read the writing of someone you are familiar with, and without seeing the authors name in the footer you’d have a pretty good guess at who they are. Working at Automattic this was pretty common for me, and it is something I loved. I often looked forward to reading content from specific people.
This is why we are so quick to identify LLM generated content, we all have access to and use the same models. We know the ‘tells’, and the signs that help us quickly identify it. The superficial takes, the verbose language, the “its not this, its that” structure of sentences. No matter the model or the ‘skills’ used, its obvious and even if you intentionally make typos, we know.
Now imagine you’re using these same chatbots to communicate with your colleagues, your audience or your community. All of these are fundamentally built through relationships, but when your LLM writes messages for you how are you expecting to form those relationships? How is your colleague expected to get to know you, or relate to you when everything you say is filtered through an LLM.
Even if AI writing becomes indistinguishable from yours, something is still lost when your words aren’t actually yours. Write imperfect sentences, have messy interactions, and express how you feel through your words.
Leave a Reply