This is a great set of takes and glad to have found you via a quote from your post in Notes.
On the issue you have found with writing with machines, have you tried iterating the writing a lot more?
I gave up drafting with GenAI many times before I realised it works well when I incorporate it more as a note taking tool as I explore a draft and iterate along. I also regularly force it to critique, challenge and question the drafts and then only once I have a draft that covers the key points and logic I’m looking for, I take it out, tear it up and rewrite the majority of it.
I notice that this process consumes much more time than my previous writing but it allows me to fragment time across many drafts as I iterate different ideas over time. Then find a block of time for the rewrite. And I find it much more interesting as a writing process (it’s still horrible and painful).
My past writing (before machines) was always a lonely struggle with feedback and perspective often only at the end or never. Yet with a more iterative approach, I find I can explore the space of my ideas while I write and have a form of feedback as I go.
But it’s still a kind fake negative to the process that seems to work best. Constantly seeking for criticism and refusing to accept that your ideas are worth writing about. Because the machines want to convince you otherwise! 😂
your approach is a good way to think of it — as a kind of mechanical assistant, though for a part of writing work that only a small proportion of people seem to seek assistance for (from anyone, human or machine).
my general rubric so far is to ask of an LLM only what you can reasonably expect it to be good at based on what it is. so i will ask it to rephrase what i write in words used by another specific audience (for e.g.) or retrieve material from a corpus.
but it does take much longer and somehow feels more tiring in different ways? at the same time, i am undoubtedly producing more stuff. idk idk. thanks for writing this long comment!
Thanks for the comment back and agree with your logic. In my work I run into a lot of GenAI-hype driven use cases. The best is to just use it for what it was designed for / does well. There are now many tools that include GenAI and do many other things and confuse the whole definition but at the core, what you describe is what it is best at.
And I do think I also spend much more time writing now than before and maybe much more time with the “writing process”. Another feature of LLMs is how engaging and entertaining they are, so I’m certain that it competes with other distractions for idle time and entertainment.
This is a great set of takes and glad to have found you via a quote from your post in Notes.
On the issue you have found with writing with machines, have you tried iterating the writing a lot more?
I gave up drafting with GenAI many times before I realised it works well when I incorporate it more as a note taking tool as I explore a draft and iterate along. I also regularly force it to critique, challenge and question the drafts and then only once I have a draft that covers the key points and logic I’m looking for, I take it out, tear it up and rewrite the majority of it.
I notice that this process consumes much more time than my previous writing but it allows me to fragment time across many drafts as I iterate different ideas over time. Then find a block of time for the rewrite. And I find it much more interesting as a writing process (it’s still horrible and painful).
My past writing (before machines) was always a lonely struggle with feedback and perspective often only at the end or never. Yet with a more iterative approach, I find I can explore the space of my ideas while I write and have a form of feedback as I go.
But it’s still a kind fake negative to the process that seems to work best. Constantly seeking for criticism and refusing to accept that your ideas are worth writing about. Because the machines want to convince you otherwise! 😂
Looking forward to your future writing!
your approach is a good way to think of it — as a kind of mechanical assistant, though for a part of writing work that only a small proportion of people seem to seek assistance for (from anyone, human or machine).
my general rubric so far is to ask of an LLM only what you can reasonably expect it to be good at based on what it is. so i will ask it to rephrase what i write in words used by another specific audience (for e.g.) or retrieve material from a corpus.
but it does take much longer and somehow feels more tiring in different ways? at the same time, i am undoubtedly producing more stuff. idk idk. thanks for writing this long comment!
Thanks for the comment back and agree with your logic. In my work I run into a lot of GenAI-hype driven use cases. The best is to just use it for what it was designed for / does well. There are now many tools that include GenAI and do many other things and confuse the whole definition but at the core, what you describe is what it is best at.
And I do think I also spend much more time writing now than before and maybe much more time with the “writing process”. Another feature of LLMs is how engaging and entertaining they are, so I’m certain that it competes with other distractions for idle time and entertainment.
Looking forward to your future posts!