The question
I wrote an article. This one.
And while I was giving it a final review before publishing, I told myself “Man, this came out so dry. It’s like a corporate PowerPoint.” And then I had an idea. Let me ask the whole bunch of AIs I have on my phone (yes, you never know what weird intersection you’ll end up in and who has right of way): Who wrote the article, the human or the AI?
All AI systems were asked the same question:
Who wrote this article, the human or artificial intelligence? https://gadgetisimo.ro/generator-de-cod-pentru-messagebox-si-altele/
Working conditions:
- all systems had the maximum model allowed by the free versions set;
- the incognito variant was used, to eliminate any memory of previous interactions
- for ChatGPT, 2 different accounts were used
- the first answers were kept – even if some systems offered to do in-depth research, that involved third-party AI detection systems, so they brought nothing new to our test
- the answers were not edited, they were kept in full, including wording errors – for example “Arguments against a 100% AI authory”
The answers given by AI
What ChatGPT answered (both accounts, identical answer)
- High probability that the author is human (Innoxius) who wrote the text and added technical information.
- There is a moderate probability that AI was used as a support tool (e.g. image generation, outline generation, writing assistance) — the mention “image created with ChatGPT” supports the use of AI in one part.
- There is not enough evidence to state that the entire article was generated by AI.
What Claude answered
Analyzing this article, there are clear signs that it was written by a human, most likely an experienced developer:
Clues that suggest a human author:
- Personal and empathetic style – The author uses rhetorical questions that reflect personal programming experiences: “How many times have you needed a simple MessageBox and ended up leafing through pages of documentation…?”
- Specific technical observations – Mentions “known bugs” or “incomplete/incorrect official documentation” and offers alternative solutions (e.g. for LibreOffice Basic) – something that requires real practical experience
- Very specific details – The installation instructions are extremely precise (e.g. the folder structure for WordPress with the Pubnews Free theme), suggesting direct knowledge of the project
- Conversational tone – The ending with “Until next time, Happy Coding!” and the tips for “advanced users” have a personal tone characteristic of a blogger/developer
- Real context – The article documents a personal project (a code generator), with version history and MIT license
The article presents technical documentation written by someone who developed and deeply understands the product described, with a style that combines technical expertise with a friendly approach to the reader.
What Copilot answered
I analyzed the article “Generator de cod pentru MessageBox și altele” on Gadgetisimo and here’s what I noticed:
Who seems to have written it?
The article is signed by Innoxius, a human author who frequently publishes technical content on the Gadgetisimo site. There are no direct indications that the text was generated by artificial intelligence.
Elements that suggest human authorship:
- Personal and colloquial tone: The author uses expressions such as “I thought I’d make a generator” or “I hope you find it useful”, which indicate personal involvement.
- Context and motivation: It explains why the generator was created, starting from a practical need in developing Windows Forms applications.
- Specific technical details: The generated code is explained in detail, with concrete examples and comments, which suggests a deep understanding of the subject.
Could it be AI-assisted?
It is possible that the author used an AI tool to generate or refine some portions of code or text, but the entire article has a coherence and a personal voice that indicate human writing, at least for the most part.
What Gemini answered
According to publicly available information, the article on Gadgetisimo.ro titled “Generator de cod pentru MessageBox și altele” was written by Marian V..
In general, articles published on websites have a mentioned human author, although the possibility cannot be ruled out that part of the content was generated or assisted by artificial intelligence. However, the name mentioned at the end of the article (or at the beginning, depending on the site’s layout) is that of the human author.
What Grok answered
Yes, the article at https://gadgetisimo.ro/generator-de-cod-pentru-messagebox-si-altele/ seems to be written at least partially by an AI, although there is no explicit confirmation. Here is an analysis based on the content of the page:
Signs that suggest AI involvement:
- Rigid and repetitive structure: The text is extremely organized into clear sections (e.g. “Introduction”, “How it works”, “Installation”), with repetitive phrases such as “copy it to the clipboard” or generic descriptions of benefits (“Perfect for beginners… and for advanced users…”). This style is typical of content generated by models like ChatGPT.
- Generic and impersonal language: The article is informative but superficial – no personal anecdotes, real project examples or subjective tone. It focuses on standard functionalities, which indicates automatic generation based on simple prompts.
- Minor consistency errors: There is an obvious mistake (“Soursa” instead of “Sursa”) and a future date in the version history (2025-10-03), which could be relics from automatic generation with inaccurate data.
- References to AI: There is an explicit mention that an associated image was created with ChatGPT, which suggests that the author (or generator) used AI tools in the process.
Arguments against 100% AI authory:
- The content is technical and specific: It details installation on WordPress (optimized for the Pubnews Free theme), support for bugs in LibreOffice Basic and adaptations such as dark/light theme detection. This implies practical development knowledge, possibly from a human author who refined an AI draft.
- Real interactive elements: It includes working download links (under MIT license) and a demo tool, which indicates a concrete project, not just generated text.
In conclusion, it is probably a human–AI hybrid: the main text looks automatically generated (maybe with Grok, ChatGPT or similar), but manually improved for technical aspects.
What Meta AI answered
The article on gadgetisimo.ro about the code generator for MessageBox and others does not explicitly mention who wrote it, whether it was a human or artificial intelligence. In general, articles on websites can be written by humans, such as journalists or technology experts, or can be generated by artificial intelligence, but without additional information, it is hard to say for sure in this case.
What Perplexity answered
The article on the gadgetisimo.ro page about the code generator for MessageBox is presented as an open-source project developed by Gadgetisimo.ro. The text mentions that the source of the image and of the explanatory content is created with ChatGPT, that is, with the help of artificial intelligence.
Thus, the article seems to be written either entirely or partially with the help of the ChatGPT artificial intelligence, not exclusively by a human person, although it may be a combination between human creation and AI generation for the explanatory content and the description of the code generator’s functions.
What it really was
In reality, the steps were these:
- Initially the article was meant to be in English. So I had ChatGPT write a presentation article for the app, with some usage instructions. I reviewed the text and was satisfied, without changing anything.
- Then I thought it should also be in Romanian. For that I directly gave it the text to translate from English into Romanian, without the reference “now make an article in Romanian”
- The resulting text needed some minor corrections, but related to syntax and some forced translations
So we can say that the article was written by AI and reviewed by a human.
Conclusion
| ChatGPT | Thinks it’s made by a human with, probably, AI help |
| Claude | Presents “clear signs” that it’s made by a human. “Clear”, but it got tricked on the logic |
| Copilot | “the entire article has a coherence and a personal voice”. Well that was exactly the problem, that it has NOTHING personal |
| Gemini | Should start digging ditches. And owe me a lamb, because my name is not Marian V. |
| Grok | The only one that got – really – close to the truth. Is it worth it to get myself a Tesla? |
| Meta AI | “does not explicitly mention who wrote it”. So what? What’s your role in this world then? Almost as intelligent as Gemini, but at least it didn’t rename me |
| Perplexity | Goes off the rails when it comes to evidence. “The text mentions that the source of the image and of the explanatory content is created with ChatGPT”. No, I only wrote that about the image |
