Reader
AI cleanup is becoming a new freelance industry.
Read time: 5 minutes
↓
 |
$5.00
Feed the rebellion.
If something I wrote saved you time, changed your mind, or made you laugh loud enough to concern nearby coworkers, you... Read more |
In my work, I hear the phrase more and more: “Let’s clean the AI slop.” This shows up in the context of landing pages, newsletters, posts, design … anything marketing touches.
Which is showing a very interesting new path … the human work is migrating.
The other day I was thinking: before AI, our access to knowledge, information, resources, tools, time, and energy was comparable to a small, private grocery shop. Where you walk in and you have only 4 types of cheese, 3 types of bread, 1 brand of tuna, Coca-Cola, but no Pepsi.
The same was true of our ideas, time, and energy: small shelf space.
Now, AI has turned that small store into a mega-market. We have an abundance of ideas, examples, hooks, visuals or directions in minutes. If you spend enough time in this mega-market and inspect all the “products”, your basket gets pretty solid.
And even though this abundance is an advantage in many ways, it is also a problem, because all of a sudden … before, with 3 types of cheese, now you have to evaluate 20 of them. And for each, you have to decide what to select within that “product segment”, and across all other product segments aka: how will your final, full basket look like.
So the bottleneck migrated from having or not having options to judging them: Is this idea good? Is the fact true? Is it distinct? Does it sound like me/client? Does all of this work together, well??
And this shift is also starting to show in the freelance market.
Fiverr reported a 641% increase over six months in searches for freelancers who can humanize AI-generated content.
So, the easy conclusion is that we are witnessing the first signs of new business opportunities:
Before: brief → copywriter → finished copy
Now: brief → AI first pass → copywriter / editor → humanization + verification + creative direction → finished copy
And that creates ANOTHER opportunity for people who can design the workflow itself: system thinkers.
This is what I mean by the human aftermarket: paid work that appears or migrates because AI took over an earlier part of the process.
I will give you another example from my high school and university days. My teachers always found sh*tloads of grammar mistakes, because I did not have the patience for the details.
I always needed somebody to grammar-correct, while I was very good at identifying and creating the larger picture idea.
That matches what I see in practice with AI.
Take writing. If you use AI regularly, you start recognising patterns at the level of:
- paragraphs
- sentence construction
- visual hierarchy
- words
- hooks
→ prompt for hooks, to proove my point:
Research current popular hooks viral creators are using on Instagram, LinkedIn and TikTok. You have to specialize for each platform and understand its algorithm. The patterns you return must be verified against 20 different viral creators. Give me 10 powerful hooks for Instagram, TikTok and LinkedIn, all unique and templatized for whatever use I need. You are not allowed to invent anything. Also report direct links to your sources. Don’t come back until you process the batch. Report only results, no commentary.
The hard part now is choosing which hook. The advantage is seeing these gaps at a much finer level than everyone else.
The more closely you understand how an output is built, the easier it becomes to see where language models still fall short.
To see the gap, some of us are taking language apart almost word by word:
- we build word banks
- deconstruct sentence rhythm
- study paragraph flow
- map tone
- identify repeated AI constructions
(And I hate this part, because it is at a micro level as grammar correction example, and you can definitely see my work leaking at this level.)
From the outside, that micro obsession can look crazy. In practice, it is a way of understanding the distance between generated output and finished work at a much finer level than “this sounds like AI slop.”
The same logic applies beyond writing.
Let’s take research:
→ prompt:
Act as a researcher on X. Find recent and evergreen academic research and other credible sources on the topic, and attach links. Summarize the findings. No hallucinations and no twisting the meaning.
Easy peasy.
The problem now is:
- you have to manually check the source and read through xx amount of articles
- you have to compare definitions and reasoning
- you have to spot what is missing from the interpretation
- you have to understand what is not comparable
Third example: design.
→ prompt:
Create a high-converting landing page for X type of business model. First, research proven high-converting landing-page structures from similar industries. Then populate the structure with knowledge from X, Y and Z files in my Notion database ZZZ, and apply the design referenced in the attachment. >
Again, easy peasy.
You put this in ChatGPT Astra or Fable 5 or 5.1 and you get incredible results. But still, you have to:
- research the buyer and understand them (if you are doing this for a client)
- clean the structure and hierarchy
- change the language into your buyers’ language and your own word bank
- make it feel like you, while using details from your buyers’ psychology, nostalgia or line of work that feel familiar to them
Etc.
Examples of projects Des and I pitched:
Example of AI-assisted pitch 1
Example of AI-assisted pitch 2
In both examples, I had to refine the AI work with my own creative direction and strategy: how to approach the idea, structure the pitch decks, and build the landing pages so they are fully personalised.
And still I managed to f*ck up slide 3 on both pitch decks, because you have to wait for 5 seconds to see the real magic.
Even though all of this was much faster to make compared to one year ago, the devil remains in the details.
Bottom line:
If you can identify where a process repeatedly needs human correction, you can improve your own workflow and build a service around the gap.
Or in other words:
You have a great opportunity to get “fck you money”, and even an opportunity to build your solo business FAST, without outsourcing what used to cost thousands.
So instead of mourning the old times, I think technology is creating a lot of new opportunities for all of us. We just need to correct the angle of seeing things.
___________________
Sandra @ Thought Readership™
| Pick the line that's most true right now. |
|
|
|
|
|
|
