Information overload is an occupational hazard of ghostwriting and editing. I’m used to plowing through decks packed with intricate charts, proprietary analyses, interview transcripts, background research, and outlines that have survived dozens of revisions. My job is to reduce them into something with enough informational density to be worth a busy executive’s time.
As a journalist, I learned to absorb large volumes of information quickly. Later, when I developed Boston Consulting Group’s content development best practices, I encouraged authors, writers, and marketers to ask questions before drafting, such as what evidence actually supports the thesis, and how much evidence does this audience need?
Those questions still hold up. But little of this has prepared me for the deluge of AI writing I now face.
I’ve received more than 80,000 words of AI-generated transcripts and content for reports that ultimately became 10,000 words. That’s a full-length book’s worth of source material. I’ve wrangled multiple versions of the same AI-iterated chapter, each different, each taking a lot of space to say what it says. Granted, much of it reflected the author’s ideas after self-interviewing, prompting, and iteration.
AI has made it almost effortless to produce an avalanche of text. Years ago, it would have taken months or years to assemble that much material. Now the bots happily drone on until someone tells them to stop. It’s a veritable case of logorrhea.
Several editors I know have started declining this kind of work or charging substantially more to work with AI-generated writing and materials. It’s challenging, but I’ve adapted instead. I use AI against AI to organize, summarize, and navigate the flood. Even then, there are days when it feels like I’m drowning in words. I thought AI was supposed to save time.
AI is also changing how I scope projects. I’m increasingly asking clients to provide synthesized materials instead of every prompt-generated draft they produced along the way. Raw interview transcripts are the exception. I still read those carefully because that’s where the author’s thinking, language, and voice actually live. I’ve started incorporating all this extra reading and thinking time into my not-to-exceed bids. Soon I’ll be charging by the gigabyte of source material.
One small suggestion for anyone using AI to develop thought leadership: slow down. Before typing a prompt, sketch an outline by hand. Or type it into the notes app on your phone. Distill the idea into a headline and a summary sentence or two. Take the time to outline an argument by hand at a granular level before asking AI to elaborate on it. GIve it a word count similar to the content needed.
Editorial collaborators don’t need every possible word a model can generate. We need the relevant thinking.