The AI revolution has a growing vacancy: someone to sort out what the AI revolution just handed in.

The Guardian reports that freelancers are increasingly repairing AI-generated work, with some describing the assignments as poorly paid, frustrating and creatively draining.

The market has numbers to go with the complaints. In an August 20 analysis, Upwork said postings seeking people to repair or finish AI output were growing 70% year over year, measured using a trailing 12-month average.

Examples included a client wanting a generic AI-made design improved, and another needing generated illustrations refined for a printed children’s book.

Upwork calls this “AI remediation,” which sounds considerably better in a budget meeting than “we need someone to rescue the thing.”

In a pleasing methodological detail, Upwork classified postings using three AI models. All three had to agree that AI output had created the need for human work.

Even counting the cleanup now has an AI department.

These postings do not establish completed contracts, earnings or net job creation. A growing repair business can coexist with a shrinking market for the work being repaired.

A study published in Management Science found a 21% relative decline in postings for writing and coding jobs vulnerable to automation during the eight months after ChatGPT launched, compared with jobs requiring more manual skills. It also found a 17% decline in image-creation postings following the arrival of image generators. The remaining automation-prone jobs were more complex and offered higher pay.

That research examines an earlier period. It cannot tell us whether today’s cleanup assignments compensate for lost opportunities. Celebrating the repair queue as an employment success would be rather premature.

For content teams, the useful question is how much work remains after the text appears. Our guide to spotting AI slop identifies problems that survive a cosmetic rewrite: unsupported citations, stale information, interchangeable examples and claims of testing with no evidence of a test.

Removing a few robotic phrases will not make an invented source exist. The sentence can sound beautifully human and still be wrong.

Readers have their own ways of objecting. As SEW reported, LinkedIn’s AI slop feedback option has been appearing more widely on comments, allowing users to flag replies they consider empty or automated. That feedback does not automatically remove a comment, but it gives the audience a button for its disappointment.

Google’s guidance on generative AI content similarly emphasizes accuracy, quality and relevance. It acknowledges useful applications for AI while warning that generating many pages without adding value may violate its scaled content abuse policy.

For a business commissioning content, this suggests a less flattering way to calculate AI savings: include the research, checking, editing and approval needed before publication. A draft awaiting four hours of repairs belongs in the work-in-progress column, however confidently the chatbot announced it was finished.

We make a related point in our SEO pricing guide: compare the cost of completing the same job, including the time your own team contributes. When a provider uses AI, ask what the efficiency buys you and who takes responsibility for checking the result.

The post Demand for people to fix AI-generated work is climbing, reports find appeared first on Search Engine Watch.

Source:: searchenginewatch.com