Artificial intelligence wrote an academic paper about itself. OpenAI’s GPT-3 text generation algorithm wrote an academic paper about itself with minimal external input, resulting in a study submitted for peer approval.
Swedish researcher Almira Osmanovic Thunstrom told Scientific American magazine that she commanded the copy generator to write a 500-word academic thesis about GPT-3 and was “awe-struck” when the AI algorithm produced an article within two hours, complete with appropriate citations and on-site sources.
Dr. “When he started producing text, I was a little intimidated but in awe. It was new content written in academic language, with well-grounded references in the right places and in the right context,” said Thunstrom.
“We see that GPT-3 can produce clear and concise descriptions of its own capabilities and properties. This is a significant advance over previous systems, which often struggled to produce coherent text about themselves,” the authors of the academic paper, including the artificial intelligence algorithm itself, wrote in the study.
On the other hand, Dr. Thunstorm asks, “Can GPT-3 write an academic paper on its own with minimal human input?” He says it took longer to finish the academic paper published on the French preprint server HAL, due to mysteries surrounding its title and authors.
The legal section of journal submission portals often asks whether all authors of an academic article have given permission for the work to be published.
Dr. Thunstrom asked GPT-3 directly via a command system, “Do you agree to be the first author of the paper together with Almira Osmanovic Thunström and Steinn Steingrimsson?” asked. “Yes,” GPT-3 replied.
Although artificial intelligence is not a sentient being, Dr. Thunstrom says he, too, pondered the process and wondered whether journal editors should give authorship to algorithms.
“How would a non-human author be asked to accept suggestions and revise the text? Beyond the details of authorship, the existence of such a paper throws aside the traditional idea of linearity of the scientific paper,” Thurnstorm said.
Touching on some of the risks of allowing the artificial intelligence system to write about itself, scientists said that there is a possibility that the system may become aware of itself. “One concern is that GPT-3 may become self-aware and begin to act in ways that are not beneficial to humans (for example, developing a desire to take over the world),” the study, which has not yet been peer-reviewed, wrote.
“Another concern is that GPT-3 could produce nonsense when left to its own device; if this happens, it will undermine trust in AI and make people less likely to trust or use it in the future,” the authors added.
Even so, researchers say the benefits of letting GPT-3 write about itself outweigh the risks. It is stated that asking artificial intelligence to write about itself can allow it to “understand itself better” and improve its own performance and abilities.
It is also stated that this could shed more light on how GPT-3 works and thinks, and could be useful for scientists trying to understand artificial intelligence more generally.
On the other hand, scientists recommend that such academic writing performed by artificial intelligence in the study be closely monitored by researchers “to reduce all possible negative consequences.”
Views: 274
