I’m HelPInG!!11
I am not an active user of so-called “AI-Assistants”, neither for coding nor writing nor research. I constantly test and evaluate those systems (for my private analyses as well as for my day job) but they don’t really integrate well with my workflow and thinking.
There are many reasons for avoiding those systems: Skill decay especially in abilities like “critical thinking” or “problem solving” comes to mind. The ecological impact of those systems is another aspect that can’t be overlooked if one cares about the world. The way those systems harness and accumulate all data and information they can get without respect or compensation for the creators while charging for their systems. All very true, important and enough to stay away from them.
But today I stumbled over an article on Heise (for the non German-speakers here: Heise is one of the biggest and most reputable tech news sites in Germany) where they evaluated Microsoft’s Copilot and the image they generated for their illustration basically says it better than I ever could:
I live with a kid and 2 cats so I am well aware of how it feels when someone randomly starts typing on your keyboard while you are trying to do something. It’s not exactly helping to say the least.
This image illustrates how I feel those systems interact with my cognitive processes: I am neither a great writer nor a great programmer nor a great analyst or thinker. I am okay at all of those things, I can get some work done. But it’s always a struggle: Programming stuff is hard and annoying. Reading stuff is hard and annoying. Writing stuff is hard and really fucking annoying. But the “hard and annoying” part is necessary for me to feel like i have a grasp on what I am doing.
A lot of stuff I write might be dumb or wrong or badly argued. But I can own all that because I put in some work thinking, reading, talking-to-myselfing and all that to reach some form of conclusion. Sometimes it’s wrong and I get to learn from people giving me input and feedback and comments. But at the point of writing I can stand behind it. I have a certain level of confidence.
“AI” assistants undermine that feeling: They are a random hand that types stuff while I write, some of it horrible, some might even be good but I never know. I never know which statistical effects brought forward a certain turn of phrase or a line of argument. They create distance between me and my thinking and my writing, they alienate the visible output of my work from my work. They alienate me from my writing.
Writing isn’t exactly thinking. Writing is writing. (And coding is a form of writing to me just with more rules.) But writing is a useful tool to help one’s thinking. Thinking towards writing something down can help me create structure in my thoughts. Can help me focus. It also forces me to actually get to some form of conclusion or result – as temporary as that result might be. And assistants disturb that process. And unlike cats or kids they’re not even cute or loveable.