Artificial intelligence (AI) has once again grabbed the eyeballs, and this time it is in the field of academics. Only recently, AI was embroiled in a controversy wherein Blake Lemoine- an engineer from Google had claimed that LaMDA had become sentient, and was immediately placed on paid sabbatical by Google citing a violation of confidentiality policies. Now, AI has added one more feather to its cap. Let’s have a detailed look at it below:
AI in academics
A Swedish researcher Almira Osmanovic Thunström at Gothenburg University instructed GPT-3 – an AI algorithm to write an academic thesis in 500 words about GPT-3 with scientific references and citations inside the text. The researcher was taken by surprise as the text started generating, since the GPT-3 was able to pen a decent introduction about itself.
Thunstrom then set out to develop a complete research paper from GPT-3 after the successful experiment and publish it in a peer-reviewed academic journal. But can a paper that is derived from a nonhuman source be published?
Thunström discussed the experiment in Scientific American, noting that the publication of GPT-3 raised a number of moral and legal concerns. Thunström wrote: All we know is that we opened a gate. We just hope a Pandora’s box wasn’t opened.
The GPT-3 did an excellent job by completing the scientific paper in just two hours. Thunström then started the process to submit the work and had to ask the consent of the algorithm to publish the same.
The reply was a Yes, Thunström wrote. I checked the box for ‘Yes,’ slightly sweaty and relaxed (had it said a no, my conscience could not have allowed me to go any further).
When the algorithm was asked whether it had any conflicts of interest, the reply was a no. Thunström wrote that the authors started to treat GPT-3 as if it were a sentient being, despite the fact that it was not.
Academic publishing may need to adapt to a future of AI-driven manuscripts, and the worth of a human researcher’s publication record may vary if something nonsentient can claim credit for some of their work, Thunström wrote.
Thunström told Insider in an email that the experiment yielded positive results in the artificial-intelligence community and that other scientists are attempting to replicate the results. Those conducting similar studies have discovered that GPT-3 can write about any subject, she claims.
Our goal was to arouse multilevel debates on the role of AI in academic publishing, Thunström explained.
