ChatGPT, the chat bot developed by OpenAI, has become the most talked about thing in the internet world in recent months. The AI tool could quickly answer whatever you ask, write articles and even poems. We’ve even recently seen it prevent someone from getting a parking ticket.
The fact that the artificial intelligence tool is so impressive, of course, is in everyone’s mind, “Will everything that humans write right now be written by artificial intelligence?” brought the question to the fore and showed that there could be a revolution in many issues from broadcasting to social media. Of course, the world’s largest internet encyclopedia, Wikipedia, wondered whether such a thing would happen. This question was answered by Jimmy Wales, the founder of the site, who spoke to the Evening Standard.
According to Wales, ChatGPT is still not quite ready to write a Wikipedia article
Wales, one of the founders of Wikipedia, talked about how this giant encyclopedia might turn out in the future, saying that he thinks a lot about whether the contents are written by artificial intelligence. The 56-year-old entrepreneur stated that people in the Wikipedia community are currently cautious because the models we currently see are not good enough, but are also curious about the possibilities that may arise.
Wales further stated that ChatGPT is still not ready for this, but he cannot predict when it will be ready. “I think we’re still a long way from saying ‘ChatGPT, please write a Wikipedia article on the Empire State Building’. However, I don’t know how far we are from that,” the entrepreneur said, adding that in the last two years, much better developments have been made than he thought.
A big problem is that the model can give incorrect information.
Wales also stated that there are still many problems with ChatGPT: “One of the problems with the current ChatGPT is what they call ‘hallucination’. I call it ‘lying. He has a tendency to make things up out of the air. This is very bad for Wikipedia. We have to be really careful about this.”
The fact that artificial intelligence tools give false information was already a hot topic. Many times before, people have presented evidence that models like ChatGPT sometimes do not provide accurate information. For example, a month ago, an economist at the University of Queensland asked ChatGPT, “What is the most cited economics article of all time,” the Evening Standard reported. he asked a question.
The chatbot found an answer to that question within seconds and said that the article was “’A Theory of Economic History’ by Dougless North and Robert Thomas, published in The Journal of Economic History in 1969.” Although this answer seems to be true, when the article was researched later, it was revealed that there was no such article. The authors and the journal were real; but such an article has never been written. This revealed that the model can sometimes give fabricated answers.
ChatGPT Saved A Dog’s Life: Vet Undetectable Disease Detected by GPT-4
In addition to these, Wales added that there are contradictions within ChatGPT itself. The entrepreneur asked the AI tool, “Has the Empire State ever been hit by a plane?” asked the question. The model, on the other hand, replied that such a thing did not happen at first, but then a plane crashed into it. This showed another problem of the technology.
How can AI models be integrated into Wikipedia in the near future?
While all this poses a problem for Wikipedia, Wales also says that artificial intelligence can offer some advantages for the site. An example is the possibility of using this technology to check if two Wikipedias are in conflict with each other. Thus, it is stated that people will not have to read both articles and this can be done automatically.
Finally, there are also concerns that artificial intelligence may be biased. In other words, an AI-powered Wikipedia can make the platform even more biased, which has already been criticized for being biased (for reasons such as the majority of its editors being white).
These explanations show that artificial intelligence-generated content will not come to Wikipedia for now, due to the flaws of the models. However, if the development of technology continues at this rate and the problems can be eliminated, such a possibility may be real.
15185
