I Asked ChatGPT: What’s the Cost of Human Workers vs. AI In 4 Popular Jobs?
Written by
Nicholas Morine
Edited by
Amen Oyiboke
Commitment to Our Readers
GOBankingRates' editorial team is committed to bringing you unbiased reviews and information. We use data-driven methodologies to evaluate financial products and services - our reviews and ratings are not influenced by advertisers. You can read more about our editorial guidelines and our products and services review methodology.
20 Years
Helping You Live Richer
Reviewed
by Experts
Trusted by
Millions of Readers
Artificial intelligence is the talk of the virtual town, with AI agents poised to replace workers in a variety of jobs, largely those involving mental labor — at least at this point.
However, perspectives differ on how likely this takeover is. Concerns range from accuracy — AI models are known to “hallucinate,” or provide fabricated information at times — to, perhaps most pressingly for market-minded individuals, overall cost structures.
Based on data from OpenAI, the Bureau of Labor Statistics and McKinsey & Co., a cost breakdown on four such positions — human labor versus agentic AI labor — follows, as outlined, ironically enough, by ChatGPT.
Customer Service Representatives
Cost: $42,830 for human labor
ChatGPT provided a baseline median annual salary of $42,830 for U.S. customer service representatives, based on 2024 BLS data.
How does that compare with AI equivalents? Using ChatGPT 4-o mini as the replacement model (running in real time through API), inference costs come to only $86 annually for the workload itself.
But citing McKinsey’s overhead estimate of $200,000 annually, the picture changes. Human CSRs are less expensive for businesses with up to about five employees in the department. Beyond that — not accounting for necessary human oversight of chatbot interactions, such as quality assurance, complaint escalation and retraining — AI agents may prove more cost-effective.
Content Writers and Authors
- Cost: $72,270 for human labor
Content writers and authors earned a median annual income of $72,270, according to the BLS. But this job could be more complicated to replace with AI.
ChatGPT calculated token costs at just $12 annually per author seat, but upkeep is substantial: A “shaper” model built to handle editorial workflows could run about $300,000 in plug-in upkeep, plus roughly $500,000 annually in model maintenance.
The verdict? “The total cost of ownership (TCO) [is] hundreds of thousands annually if scaled to enterprise; per-seat usage is cheap, but the infrastructure is not,” the model noted.
A writer’s room would need about seven writers or more to consider full deployment — and even then, at least one or two human overseers would be required to curate and correct workflows.
Editors
- Cost: $75,260 for human labor
The median U.S. salary for editors is $75,260. An AI editor may be useful for spelling and grammar, as ChatGPT acknowledged, but for contextual understanding, fact-checking and appropriate tone, the results are less reliable.
Based on McKinsey’s shaper-level archetype and OpenAI’s pricing, an AI editor would cost about $200,000 annually in TCO. Just $41 of that comes from token use.
Graphic Designers
- Cost: $61,300 for human labor
Graphic designers, facing heavy competition from generative AI tools, earn a median salary of $61,300. The AI equivalent, however, would set a business back about $200,000 annually in TCO, according to ChatGPT. Only about $452 of that is token spend, with the rest tied to infrastructure needs.
The Takeaway
“The raw token costs are tiny — tens or hundreds of dollars per seat per year. But McKinsey’s recurring platform and governance costs ($200,000-$500,000) dominate in every role once you move from experimentation to production,” ChatGPT explained.
“That’s why ‘AI is cheaper’ claims are misleading unless you factor in full lifecycle costs and human oversight,” it concluded.
More From GOBankingRates
