On writing with AI
If you've not been living under a rock for the past year, it's very prominent how much of the news cycle revolves around AI these days. It's surely exciting but also a scary thing to witness as these LLMs get pretty good at things like knowledge work, programming, etc.
Amongst the people I've interacted with, some are very much pro-AI while there are others who abhore the entire infrastructure behind it. I'm in neither camp and prefer to keep a more pragmatic stance on the subject. Of course, there are multiple ethical concerns regarding AI usage: how these datacenters are causing water, land shortages in addition to increasing global temperatures, and how it's okay for the AI companies to train their models on copyrighted material but if a human does it, it's a criminal offense[1]. I also hate how frontier AI companies like OpenAI and Anthropic have a huge power on dictating how AI should be regulated (a major reason being that they're simply afraid of open-weight models coming out of China).
Leaving all of these concerns aside, I do see how useful LLMs can be when it comes to doing a quick online search, topical research or coding. Just last month I attended a talk by Prof. John Martinis, where he mentioned that he uses LLMs as a "super search engine" when researching about a new topic. These models however, hallucinate a lot so if the model makes a claim, he would ask to cite the particular paper from literature where the model made the claim from. I think this is a very good use of AI where one doesn't delegate all of their thinking to these models.
In my daily work as a PhD student, I do use AI as a search engine most of the time, or to do knowledge research. These days, I mostly use it for programming (although I don't use the agentic way of coding, which for me sucks the joy of using a computer). I do however, write my notes by myself. Writing is just as important as thinking, since one uses their own thoughts to do the writing. If you delegate this part to AI as well, then what exactly is your brain for?
Currently, I'm teaching a lab course for bachelor students and sadly see them using AI in their writing. It's not always easy to tell but some hints of Claude is quite easy to recognize i.e. "it's not just X, it's Y", the liberal use of em-dashes, etc. At this point, I'd rather read text with typos than the very generic and soulless way of writing that is made with LLMs.
It saddens me that in addition to a majority of blogposts that I come across online being written with AI, I'm also seeing in real time, how much students are using AI in their education. Please don't be a meat proxy.
In my work and personal life, if I see a hint of AI in any writing (or creative work for that matter i.e. music, video, etc.), I make a deliberate choice not to peruse it. I understand if one uses AI for proof-reading, checking grammar, correcting for typos, etc. but if you completely delegate the task to the LLMs to vibe-write it for you, then I don't want to waste my time reading slop.
-
The irony of that statement is baffling to me. ↩