Can you distinguish between what a human writes and what artificial intelligence produces?

You are browsing the Internet, and an article or text catches your attention and you read it. But have you ever wondered who its real author is? Is he a human or a robot?

A few years ago, this question seemed like an assumption, but today it has become part of the daily reading experience on the Internet.

A new study titled "How much AI-written content is on the Internet?" examined The Pew Research Center in the United States conducted a survey of hundreds of thousands of texts and materials spread on the Internet, to measure the frequency with which signs appear indicating that some content was created or modified using artificial intelligence.

The study, published on August 20, analyzed about 490,000 English-language web pages and found that about 10 percent of all pages published in the July 2026 sample showed strong indications of being created with artificial intelligence, while the percentage exceeded a third among pages published after the launch of GPT chat. This reveals the extent to which artificial intelligence has become an essential feature of the content creation process today.

But can we really distinguish between what a human writes and what an artificial intelligence produces? Who uses these technologies the most? More importantly, what about Arabic content? Is the use of AI in its production moving at the same pace as we are seeing in English-language content?

AI does not necessarily leave a clear sign that this text is machine-generated. Instead, researchers look for a range of linguistic and stylistic indicators such as the words used, the way sentences are structured, punctuation, and recurring writing patterns.

Hoda Boumar, an expert in natural language processing and machine learning, and an associate professor at Carnegie Mellon University in Doha, explains her vision on how to distinguish between content generated by artificial intelligence and content written by humans without automated intervention. She tells BBC News Arabic: “Theoretically, it is possible to detect texts produced by artificial intelligence, but the task becomes more difficult as models develop. In English, the tools are more advanced, while in Arabic they are still limited due to the absence of large reference datasets on the one hand, and the diversity of Arabic writing styles such as classical Arabic.” Written vernacular, on the other hand, and, moreover, machine-generated texts may appear more perfect than human writing, making the distinction between them and human texts more complex.”

According to the study, the use of the dash mark (-) has almost doubled since 2023, and the use of certain words and phrases that researchers have identified as features associated with artificial intelligence-generated texts has also increased.

But the study confirms that one mark alone cannot prove that a text was written by artificial intelligence. The idea is to collect a number of indicators to estimate the possibility of the presence of artificial intelligence.

Hoda Bu Omar believes that the Arabic language faces more challenges with artificial intelligence than other languages, “due to the small amount of digital data available in Arabic needed to train artificial intelligence systems compared to English.”

The great diversity between classical and dialects, the complexity of word structure, and the lack of reliable data, especially in Arabic dialects, are factors that make artificial intelligence systems understand and deal with English better in many cases, according to Hoda Boumar, who worked with a number of her colleagues on the MADAR project, which is (applications and resources that support multiple Arabic dialects) with the aim of building parallel resources and blogs that cover dozens of Arabic dialects.

On the other hand, Noha Miller, a professor at the University of Sharjah and a researcher specializing in Arab media and digital journalism, told BBC News Arabia: “It is true that artificial intelligence’s understanding of the Arabic language is improving, but the challenge is not limited to language only, but also includes culture. It may produce linguistically sound text, but it reflects Western references more than Arabic. Not to mention that the Arab systems themselves face difficulty in understanding some dialects due to the great cultural and linguistic diversity in the Arab world.”

Nizar Habash - a professor of computer science at New York University Abu Dhabi and a researcher specializing in artificial intelligence and digital processing of the Arabic language - believes that the challenge and the impact of artificial intelligence on the Arabic language “is the duality between dialects and classical Arabic, because the latter is not the mother tongue that people acquire automatically, and mastering writing in it requires a lot of training and practice, and since reading and evaluating any text is easier than writing it from scratch, here comes the role of generative artificial intelligence tools that provide tremendous assistance, which is lowering the threshold for writing in classical Arabic.” And allowing a greater number of people to express their thoughts in a sound language, and this, I believe, is a true gift for the Arabic language.”

After the launch of GBT Chat, a number of Arabic language users - like other language users - have come to rely on artificial intelligence in drafting some emails, text messages, and even in translation instead of traditional translation tools and creating content from scratch.

Although there are no statistics on users of the Arabic language, Professor Habash believes that the uses of artificial intelligence do not differ much from other languages, especially in correspondence, refining style, correcting errors, translation, preparing official documents, summarizing texts, and generating initial drafts of content.

Habash believes that its usefulness in Arabic may be greater, because it helps the user move from an idea or spoken dialect to an eloquent and clear text.

So, can we say that using an artificial intelligence technology makes the user more adept at formulating words, or does it reduce the ability to develop and express his own ideas?

The danger is real, but it depends on the method of use, according to Professor Nizar: “Artificial intelligence is just a tool that can be used as an alternative to thinking and discussing, testing, and expanding our ideas. Whoever uses it to produce ideas that he does not understand may seem eloquent without having real content, and whoever uses it to review and challenge his ideas may become more accurate and creative, and this problem preceded artificial intelligence.”

This is perhaps the most important question in the whole story.

The problem is not only that machines have become able to imitate human writing, but rather what happens when humans themselves begin to write in the way that machines suggest, and can artificial intelligence-generated writing affect the way journalists and writers write in Arabic?

Professor Noha Miller says that the possibility of generating artificial Arabic is very real.

But she confirms, "We need long-term studies to prove this possibility. In any case, the current systems tend to generate content in the classical language more than trying to understand all dialects. Most likely, this may lead to the emergence of a style of digital writing that leans more toward classical Arabic. In addition, existing studies indicate that a number of Arab journalists use artificial intelligence in translation, summarization, and editing, and all of this affects the journalistic style."

But at the same time, Miller believes that the relationship is interactive, meaning that “the artificial intelligence learns from the journalist’s language and the journalist in turn uses the artificial intelligence to write. However, excessive reliance on artificial intelligence may lead to repetition or similarity of linguistic styles.”

So, will the impact of artificial intelligence on the future of Arabic writing be positive or negative?

Professor Nizar Habash says that relying on artificial intelligence is a mixture of opportunity and anxiety, with a clear tendency towards positivity. In his opinion, artificial intelligence represents a very important development in terms of encouraging Arabs to use Arabic instead of switching to English or French, but at the same time it raises serious concerns about the decline in creativity, basic knowledge, and the ability to retain information and think independently.

In other words, Habash believes that the positive future does not lie in artificial intelligence taking over the writing for us, but in helping us write and think better.