Last week when I was writing my blog, I decided that I would try and use A.I to improve the piece. Along with the firefox browser comes access to Anthropic Claude. I asked the AI to critique the piece, then suggest things to work on and to rewrite the blog with an eye for fixing the “flaws” that were in the work.
Having done that I was told I have inconsistent tone. Casual mixed with Gravitas. That seemed on brand for me. When I have a conversation often it goes from silly to serious and back again. This might not be effective in writing so I’ve been reflecting on that.
Next it critiqued my organizational flow, but had no improvements that it made rewriting the piece. It seems too many people online follow some sort of step by step essay writing process and because AI trains on that sort of data, it expected more of the same. I disagree, I think to hold people’s attention it’s important to give things a general form but to allow thoughts to break up the “landscape” of your writing.
It critiqued that I didn’t make contemporary problems the start of my piece. And sure if I was writing a recipe (not a modern website recipe that tells about your history with the food you’re preparing) I’d want to know exactly what problem was being addressed and the solution would be spelled out. But I wasn’t writing that. Though I was addressing an issue, I really wanted to allow people to imagine themselves as heroes from different eras and see that people have been facing similar problems for a long time. And that we can work through those issues.
The thing it mentioned that seems most valid is that I’ll include tangential details.
Next it said I gave a surface level description in things and that I needed to elaborate more.
So for all it’s critiques I found that maybe 2 of 5 had merit. Is that enough to make this a useful tool for writing? I don’t think so. I think your time might be better spent creating a checklist of things to look out for and use that. And the reason I think that’s better is because I had AI rewrite my work twice.
The first version it said it corrected grammatical errors while introducing new ones, and in both versions the AI ended up sounding like a self-help salesman.
Which made me worry about my writing. Do I parrot too much I know better than you sentences and vibes? Am I really writing something worth telling the world or am I telling people things they already know and make their eyes roll in their head as they read it?
The thought occurred to me that perhaps a good use of AI is to see how original your approach is, or perhaps to cliche check. But I find that to make a well done product AI is a very rough tool.
I was thinking about this when I was asked this week how I see A.I affecting the business I’m in. In my particular case it’s not affecting the business as much as others, and I think part of that is that construction equipment by it’s nature of being rugged and long lasting has to be one of the last adapters of tech. The tech at that point should be well understood so that no new safety hazards are introduced and that less educated people can easily use it.
I wish I could say that I feel that AI is ready to revolutionize the world in a positive manor but right now it feels like a promising first draft of a short story more than a finished work.
My advice for people who use it is that you should use it as a planning tool and not a “doing” tool because it’s not ready to “do” nearly anything that it does well. It’s passable at a lot but as you get more specific it gets less useful.
I don’t know how you feel about AI but I hope you think before you use it.
Chat GPT?
Good read 👏🏽