I drafted a proposal last night. 20 pages, 5 questions, some of them running to 3 or 4 pages each. Also the budget calculations and detailed breakdowns. A huge piece of work, which took me just 1 day: 8 hours, start to finish, from studying the tender brief to a finished document ready to share with colleagues for final review.
If I’d written it entirely myself, every word from scratch, it would have taken me 4 days. So on paper this is a clear win: I saved 3 days on something which is actually unpaid work. You don’t get paid to write proposals, you only get paid for the delivery if you win the contract. So any time I can claw back from this part of the process is time I get to spend on the parts only I can do (and that I know I do well): the facilitation, the relational work, being in the room with people.
But it feels weird.
(When I say “I used AI to write it” I mean I did the thinking first. I mapped the requirements from the brief, made a plan, checked the timeline, designed the activities we’d propose, worked out how many people we’d need, made sure it would fit into the available budget. That was all me. Then I brain-dumped at Claude and got it to draft the writing based on my thinking, and then I went through my usual process of reviewing, editing and fine-tuning until it said what I meant.)
I finished the proposal and shared it with my colleagues. I said, “Not gonna lie I’ve used AI a lot on this one because it’s so big and it would have taken me 4 days otherwise to write this much from scratch, but as a result I find it really hard to judge whether it’s actually good. Hope it’s good enough and it makes sense and answers the questions.” – and colleagues who reviewed validated that it’s a good proposal.
So weird though.
When I write something entirely myself, every word mine, I know when it’s done that it’s good. I can read it back and think, yes, this is solid, this follows my thinking, this is structured and coherent. I know that because I built it myself, sentence by sentence, and I understand exactly how and why it holds together. With this proposal, I don’t have that. It’s got everything in it. It definitely answers the questions. But I don’t have that same deep, in-my-bones knowing of the value of it, because I didn’t forge it myself word by word.
I’ve thought carefully about my use of AI. I have even included a statement in my policies and ways of working document.
I use an AI assistant (Claude, by Anthropic) as a professional thinking and writing tool. In practice, this means using it to structure complex reports, draft and refine documents, work through strategic or conceptual problems, sense-check my thinking, and support the kind of sound-boarding that colleagues in an organisation might otherwise provide. I do not use AI to generate content that requires professional judgement or independent decision-making, replace engagement with clients or communities, or produce outputs that I have not carefully reviewed and taken responsibility for.
I’ve turned off model training in my account settings, so the information I input isn’t used to train Anthropic’s AI models. I never input named individuals’ personal or sensitive data.
I have made a considered decision to use AI, weighing up the benefits against the environmental costs. AI supports me to deliver quality work more efficiently as a solo consultant. I am mindful not to be wasteful in how I use it. I think of this in similar terms to other pragmatic choices in my work, like driving to meetings or printing materials: real costs I need to be thoughtful about, but that support the impact I am creating. The work I deliver, helping public services improve their relational practice and make space for community voice, has value too, and using tools that help me do that well is a trade-off I have weighed up.
I keep circling back to the same justification: if I don’t spend my limited time on the things that AI can help accelerate, I get more time for the things only I can do. It helps me stay competitive as a freelance consultant, and it’s a trade-off I’ve settled for.
The thing is, I’m good at writing. It takes me a long time, I think maybe longer than other people (though I’m not sure, I mean no doubt it varies). It definitely takes a lot longer than Claude! But when I write something, I feel proud of it. It’s mine, I made this, I can point to the choices I made and why. When I write something with AI, even after all the editing and shaping I put into it, I don’t feel that same pride. It’s more like something I had to produce and get out the door. Useful, necessary, but not something I’d hold up and say, “look what I made.” That’s a bit sad, and it feels a bit empty. Some of the satisfaction of writing is tied up in the slow, effortful part of it, the part I’m trading away for speed for work reasons. I miss it.
I don’t think the answer is to stop using AI for work writing. The time saved is real, and it goes straight back into the kind of work that actually needs me, specifically, in a room, with people. But there’s a difference between “this saved me time” and “this is something I am proud of”.
The answer is maybe to start writing again, and make things not just because I need to do them efficiently, but because I want to. However long they take.
P.S. This is the first thing I’ve properly written this year. (My weeknotes don’t count.)
Photo by Planet Volumes on Unsplash