Earlier today I published a short series of posts on some ideas I had for using language models (e.g. ChatGPT and Claude) to help support academic writing.
I didn’t plan to write a series of posts. I initially had the idea to test Claude’s capability as a peer reviewer, and as I was finishing up that post, I started wondering how journals would use this technology (if they aren’t already). And as I was preparing to publish that post, I was thinking about how authors – especially novices – could use LLMs for their academic writing.
To keep things neat, here’s the collection of those short pieces on using generative AI at various points in the academic writing for publication process.
- Claude, you are an expert peer-reviewer…
- Journals should be experimenting with LLMs in their editorial workflow.
- Claude, help me draft the outline of my academic paper.
I’d say that I’m fairly confident and relatively experienced in using generative AI for a range of different activities, and I still find myself being impressed with the range of capabilities they demonstrate. This little series of experiments has convinced me further of the need to invest more time in understanding how they work, and to use to increase our capacity to learn.
I think we’re not even close to appreciating the magnitude of the impact that generative AI is going to have on society.