Can artificial intelligence like ChatGPT replace content writers? In 2026, a more useful answer won’t be a simple “yes” or “no,” but rather repetitive writing tasks will become more easily automated, while original research, verification, experience, editorial decisions, and accountability still need to be addressed seriously. AI can speed up the production process, but incorrect or non-value-added content can be detrimental to users and business performance.
This article updates a personal experiment by a Bloggerpi author that began in December 2022 and involved niche travel content. Google Analytics and Search Console data, and 11 original screenshots are retained for historical research purposes. The results should not be interpreted as a randomized controlled trial or a guarantee that AI will produce similar growth on other websites.
Summary of the 2022–2023 ChatGPT experiment and its limitations
| Author-reported indicators | Old content optimization | New AI-assisted content |
|---|---|---|
| Comparison period | December 2022 vs January 2023 | December 2022 vs January 2023 |
| Traffic changes (reported (original) | +97.4% | +39.7% |
| Avg. Session Duration | +13% | +8.1% |
| Avg. Time on Page | +17.2% | +38.2% |
| Safe interpretation | Observational results, many confounding factors | Does not prove AI is the sole cause |
These numbers come from internal reports and screenshots described in the original manuscript. Without a control group, keyword splits, and seasonality analysis, we can’t attribute the entire increase to ChatGPT. Read data-driven marketing and digital marketing analytics to understand the importance of measurement design.
1. Initial experiment goals: speed, accuracy, and organic performance
The authors tested whether AI could help generate new travel articles, edit old content, generate natural language, and bring URLs into search results. For some new content, AI contribution is said to reach around 90%, but humans still determine keywords, photos, sources, fact-checking, and editing.
For SEO content writers and SEO writing skills, this experience demonstrates why source evaluation skills remain crucial. According to the authors, ChatGPT sometimes made mistakes regarding years, distances, area sizes, and historical data during initial testing, so verification with official documents is still necessary.
2. What will change in AI capabilities in 2026?
AI can now assist with outlining, search intent classification, source comparison, data processing, and style revision with greater flexibility than early prototypes. Some tools also have access to web search and citations, as discussed in ChatGPT Search and AI SearchHowever, models can make factual errors, fabricate sources, or omit context. Not all tools use the same data or capabilities.
If the target is SEO-friendly articlesAI is better suited as an assistant following editorial briefs than a replacement for the entire research process. Responsibility for claims remains with the website owner.
3. Google doesn’t ban AI, but cracks down on spam content
Google Search Central guidelines state that AI can be useful for research and content structure, but mass page creation without added value may violate policies. scaled content abuseThe policy doesn’t depend on whether the content is written by humans, automated, or a combination of both. Therefore, keyword stuffing strategies and publishing thousands of thin articles aren’t safe just because the text has been paraphrased.
To distinguish pages that simply present generic information from unique experiences, study commodity vs non-commodity content and concept Google E-E-A-T. Take note Google spam update March 2024 as well as Google algorithm update when comparing old SEO practices with modern Google policies.
4. AI-assisted content writer workflow without loss of quality
Step one: do your research and set an editorial brief
Determine the audience, key questions, potential misunderstandings, terms that need definition, and the type of evidence needed. content planner to keep the workflow clear and ensure new content doesn’t duplicate existing pages.
Step two: produce a source-based draft
Request an outline or comparison table using verified materials. All figures, dates, legal statements, products, and experience examples must come from a traceable source. Don’t ask models to claim to have visited a destination or used a product if they haven’t.
Step three: edit as the person responsible for publication
Check that the main answer appears first, the examples are relevant, the structure is not too repetitive, and there are no sections that simply repeat keywords. on-page SEO, internal link validation, and evaluation technical SEO before publishing. AI-generated content must still provide value to real readers.
5. How to test whether AI improves productivity?
The author’s previous experiments showed that the production of 1-2 pieces of content in 2-4 hours could be increased to around 4-5 pieces of content per day with the help of ChatGPT, and estimated savings based on a hypothetical cost of Rp50,000 per piece of content. This is a personal calculation from 2023, not a standard content service price or ROI for 2026.
To test more effectively, create two similar topic groups, controlling for length, keyword type, and editor review. Measure the total time, including fact-checking, error rates, revision loops, and post-publication impact. The tool price should not equate to the total production cost if the QA process is lengthy.
6. Measure SEO content responsibly
Use Search Console for impressions, clicks, queries, and landing pages; use GA4 for engagement and key events; then evaluate in the dashboard. Differentiate the impact of content refreshes, technical migrations, keyword changes, and Google updates. Connecting SEO performance with segmented historical data is more useful than comparing two months without context.
Analysis can be run in Python/Jupyter, as in the experiences of belajar data analyst and WSL and Jupyter. However, a coincidental change in traffic following the publication of AI content does not automatically constitute a causal relationship.
7. Will AI replace all writers?
Jobs that simply convert generic briefs into generic text are more easily replaced. On the other hand, interviewing, observation, research, experimentation, reporting, critical editing, and product understanding skills remain valuable. Teams must also consider copyright, data security, and transparency in collaboration. Using crawling data to gather sources does not justify unauthorized data collection.
Archive of 11 images and original experiment results from 2022–2023
The author’s initial graphs and screenshots remain available, along with sample tour text, Analytics results, Search Console results, and index testing. All images below represent historical evidence of individual experiments, not current industry benchmark data.
See all original experimental images and graphs
FAQ: artificial intelligence, ChatGPT and content writer jobs
1. Can AI replace content writers?
It can automate some tasks, but research, validation, and editorial accountability are still important.
2. Does Google ban AI articles?
Not automatically; Google bans scaled content abuse that doesn’t provide value.
3. Can AI content be indexed by Google?
Yes, if it meets crawling and quality requirements, with no guarantee of ranking.
4. What are the weaknesses of AI-generated content?
Potential for factual errors, inappropriate references, generic content, and repetition.
5. Are the results of the Bloggerpi experiment applicable to all websites?
No; observational data from a specific niche and period.
6. Is the 97.4% increase in traffic definitely due to ChatGPT?
Cannot be concluded without controlling for other factors.
7. What is the editor’s role after using AI?
Validate sources, assess experience, improve structure, and be accountable for results.
8. What is scaled content abuse?
Large-scale content production to manipulate rankings without any useful value for users.
9. How to maintain the originality of the content?
Add research, data, observations, and examples that you actually have.
10. Is AI always cheaper than writers?
Not necessarily when verification, licensing and repair costs are taken into account.
11. How to measure the success of AI for SEO?
Compare quality, uptime, traffic, engagement and conversions with reasonable controls.
12. What writing skills are most in demand?
Research, interviewing, editing, SEO, business savvy, and critical appraisal.
External references
- Google — generative AI content guide
- Google — scaled content abuse policy
- Google — people-first content
- Google Analytics — GA4
