AI Helps to Eliminate Erroneous Wikipedia Sources
Tired of broken and erroneous Wikipedia links? This might be a solution.
According to a new study, AI can help eliminate inaccurate or incomplete reference lists in Wikipedia entries.
This is made possible thanks to London-based company Samaya AI who developed a neural-network-powered system called SIDE. Its goal is to analyze if Wikipedia references support the embedded claims.
To carry out their study, they trained the machine to recognize good references with the help of existing featured Wikipedia articles, both promoted and those which receive a lot of attention from editors and moderators.
As a result, it was able to identify content that led to subpar references through its verification system. The machine is also capable of scanning the Internet for reputable sources, and rank options to replace bad citations.
Petroni said that together with his colleagues he used the program to suggest references for featured Wikipedia articles that he had not seen before.
“In nearly 50% of cases, SIDE’s top choice for a reference was already cited in the article. For the others, it found alternative references,” the reports read.
At the same time, a group of Wikipedia users who reviewed these results said that 21% preferred the citations found by the AI, 10% opted for the existing citations. Meanwhile, 39% did not have a clear preference.
While the tool could come in handy for some editors and moderators, other scientists still underscore that in the end it’s all about what the community decided.
“The system could be useful in flagging those potentially-not-fitting citations. But then again, the question really is what the Wikipedia community would find the most useful,” says Aleksandra Urman, a computational communication scientist at the University of Zurich, Switzerland.
She adds that the Wikipedia users who tested the SIDE system were twice as likely to prefer neither of the references, which means that “in these cases, they would still go and search for the relevant citation online,” she added.
Previously, GN Crypto begged the question of whether AI makes you say things you never said?
