High pressure to publish
There are various factors influencing the appearance of unreliable scientific publications:
- The rise of the internet and artificial intelligence (AI);
- The introduction of open access publishing, resulting in the emergence of various less reliable scientific journals;
- The high pressure on scientists to produce many publications.
These changes can lead scientists to be tempted to falsify or influence results, write fake articles, or add their name to an article they didn't contribute to. There are organisations that offer these services to scientists and make money from them. An example of this is the so-called "paper mills".
Research paper mills are commercial, unofficial, and sometimes illegal organisations that produce and sell manuscripts that are not genuine but appear to be real research (1). Through such a paper mill, a researcher can buy an article and pretend it's real research, submitting it to a scientific journal. It happens that such fake articles are also accepted and published.
One of the consequences of fake articles is the creation of "zombie literature." This happens when false research results are used in other research. Sometimes it's later discovered that a published article is fake or that data has been manipulated, and in that case, the article is often retracted: it is taken offline or it is noted that the article contains incorrect information. But often, by that time, the article has already been used by other researchers, and there isn't always enough communication about the fraud in the earlier article. This creates zombie literature: research that has actually been retracted, but whose conclusions still circulate and resonate in other research. (2)
Correcting and controlling
Unreliable research results don't always arise intentionally. For example, there's biomedical research where experiments have been conducted with other types of cells than indicated. Often this isn't the intention of the researchers but is due to mistakes. Errors in research are sometimes discovered afterwards, but it's not always easy for scientists to correct these mistakes. Willem Halffman also investigates how this could be improved. This would already help to make scientific literature more reliable.
Furthermore, William's research shows that new rules and checks at journals can help detect articles written by artificial intelligence. Because while artificial intelligence can do a lot, it doesn't yet produce exactly the same texts as we humans do.
What can we still trust?
Scientists do their best to publish as correct information as possible. And the vast majority of scientific publications are fortunately carefully and correctly produced. But when people are under extreme pressure, they may find it more difficult to act honestly. In China, bonuses are sometimes paid to scientists who successfully publish. These bonuses can amount to hundreds of thousands of euros. Then it quickly becomes tempting to commit fraud. What would you do if you could earn so much money with a fake publication?