3 Tips for Working with AI Content
While on a mindless Facebook scroll recently, I saw a post that said people were surprised that they should fact-check Chat GPT. Oh my.
The person also said, I now realize that people probably weren’t fact-checking Facebook or X either. Oh my.
This is not going to be a rant about AI and how it’s going to destroy civilization. I can’t speak to that, I have no idea what may be coming down the AI pike. What I do know about AI, so far, is this:
AI doesn’t know everything.
AI knows what we know. Whatever our science is, our laws, our history – that is what it knows. It uses everything it has access to and/or was trained on, then spits out an answer to what you’ve asked, using content that was already created, not all of it true or accurate.
AI isn’t you.
It doesn’t know you. It has no idea what you’ve experienced, how you actually felt about it, what it’s like to wake up in your body. Anything that it writes is not you.
AI is meant to support you in your work.
Is it useful? Yes, for sure. I have found it to be great for brainstorming ideas for social media posts, setting up a framework for a proposal and working through some minor PHP coding to fix an error on a site.
For all of these reason, and that AI has been known to hallucinate and/or make things up, the first thing you should do with any AI generated content is:
1. Proofread.
Proofreading is a part of any writing, whether it’s a research paper, novel, or a blog post. At it’s simplest, proofreading is simply about spelling and grammar. The other reason you proofread is to edit your writing to make it sound better, to expand on an idea, remove things that don’t seem to fit now that you’re reading it and to ensure you’ve properly credited work that isn’t yours.
It’s no different with AI. You may not have to make a lot of spelling or grammar edits (ok, so maybe the EM dash though I personally never use them) but you’ll definitely want to check on a few other elements of what it wrote.
Take note of any info that sounds factual.
Proofread to find anything that seems to be factual: dates, names, places, numbers (34%, 5,000 people), etc. This should also include book titles, research papers, events, etc.)
Does it make sense?
AI can put things together but, particularly in fiction, what it strings together may not make total sense. It takes ideas that it finds and can generate some decent copy but it does not have human experience or emotion. It doesn’t necessarily understand how two scenes do or don’t fit together and how to give it emotional resonance and meaning.
Voice/Tone
Certainly you can give AI suggestions on how to make what it gives you sound – friendly, humorous, serious, etc. But you will still want to make sure that it fits with how you (or your client) actually sound and that is also works for who/what you’re writing for. Writing a job proposal is different than a blog post.
2. Verify your facts.
I often say in jest that, if its on the internet, it has to be true. That statement is, most assuredly, NOT true, and to repeat, I am joking when I say it.
All of the things you proofread for in step one? It is time to verify that those thing are, indeed, true from multiple, reliable sources.
If AI gives you a research paper, a study, or an author as a reference make sure that it truly exists.
Specific numbers, dates, cite research papers, specific names? Look. Them. Up.
As I mentioned, it does not always get things right. It may not necessarily be malicious but, in this instance, there is a very human aspect to AI: it makes mistakes. Makes sure what you’re sharing is accurate before you share to avoid situations such as fake books on a summer reading list. (fake books on book list – https://apnews.com/article/fake-book-list-ai-newspaper-summer-reading-fcdf454a5b467dad3adfed6ca1a224d2) and this (rfk fake studies – https://api.politifact.com/article/2025/may/30/MAHA-report-AI-fake-citations/).
Yes, it’s extra effort but AI still saved you time even with this additional effort and, as I said, it is meant to support you, not replace you.
3. Edit. Edit. Edit.
AI generously spared you the dreaded blank page and coming up with the writing on your own. You now you have raw material to work with. The shitty first draft, as it were, and so now, it’s time to edit.
Why should you edit?
There are a number of reasons why it’s necessary to edit our work, and I don’t think that changes regardless of whether AI wrote the text or not. I will start with the one that I feel is the most important one and most specific to AI writing.
Your Take: What do you actually think?
The writing that AI gave you is essentially a summary of what is already available on the internet and/or what someone else said. You may agree with what you’ve been provided but make sure of that. I have commented on a social post because an article someone shared, based on the title, sounded like an opinion they would hold. In reality, the post was satire and compltetly made fun of what that person believed in. Clearly, said person did not read what they posted.
As you read, anytime you think well, that could be a way to do it or say it – make a note and rewrite it. That is YOUR voice coming through. Your HUMAN voice. Let’s hear it.
Voice/Tone
This can also be tweaked through additional rounds of prompts but frankly, I don’t want to hear from AI. I want to hear from you. When you speak, are you serious? Are your IRL conversations peppered with movie lines or song lyrics? What do you sound like? Now take that and work it into your writing, to make it yours.
Super Excited Use of Adjectives
Oh my, does AI love to be excited, keen on, jumping at the chance, etc. It seems to be really, super, excited about all the things. I have a more pared back speaking and writing style. I find a ton of adjectives feels false somehow. Of course I use adjectives but just not 50 of them at a time.
Brevity
Trim it. A lot of it, just like you would trim your own writing. AI can be very wordy so just because AI wrote it, doesn’t mean you have to use it nor does it mean that it’s good. Again, AI is using content that already exists and not only are you using someone else’s work to generate yours, if you use exactly what AI gives you then you are using AI’s voice and not your own.
One of the first things I used Chat GPT for was to generate a list of 8 items about dementia for a social media post. Each item description was 3 sentences long and sounded repetitive. I wanted each bullet point to fit on an Instagram post. This work was also for a client and so I had to work with it to get it to sound like she was the one writing the copy.
In summary:
Since AI doesn’t know everything and will make things up, proofread and check your facts.
AI isn’t you (or your client) so edit and rewrite to give the work the appropriate voice/tone and give it some humanity.
AI’s best role is in support of your work – don’t allow yourself to be replaced so easily.