“What is the future of thought and reasoning that we want to preserve? If children are already being born in the midst of an excessive amount of screens, preferring to move on the screen to holding a pencil or a pen, we need to try to ensure what will become the new standard of reasoning”. It was with this warning that computer scientist Nina da Hora began her participation in the Generative Artificial Intelligence: Challenging the Limits of Reality and Creativity in Education and Research panel, this Monday afternoon (22), the first day of the WRNP. 

The conversation about the use of tools like ChatGPT in education and academic research was mediated by the leader of RNP's Data Science and Intelligence Technical Committee, Daniela Brauner. PUC-SP professor Diogo Cortiz and associate professor of the Department of Computer Science at Purdue University (Indiana, USA) Bruno Ribeiro completed the table. Nina da Hora participated remotely. 

For Diogo Cortiz, generative artificial intelligences create a new paradigm in society, constantly transforming the relationship between work and technology. “ChatGPT was trained to give not just any answer, but an answer that would convince humans. ChatGPT will never say 'I don't know'. It will always generate convincing answers – not necessarily correct, but that are plausible for humans to understand as a good answer”, explained the PUC-SP professor. 

Bruno Ribeiro made it clear that he sees artificial intelligence positively, despite its risks. “When we created the web, we automatically thought it would democratize information. But if you don't know the keywords in the search browser, you won't find the information. With ChatGPT, it's different. It will be fantastic that people don't need to know exactly the keywords to look for the information they need”, said the researcher.

Although she has issued warnings, Nina da Hora believes that humans will not be replaced by machines in the near future. “I am a little skeptical of this conclusion [that we will be replaced by machines] because the creation process cannot be visualized only in the result”.

Generative artificial intelligence and fake news

The use of AI to produce fake news was also highlighted in the panel discussion. In the view of the Purdue University professor, ChatGPT does not amplify the problems related to this issue so much. “There are already enough humans to create fake news. The big problem with fake news is its dissemination, not its production”, said Bruno Ribeiro. 

Nina da Hora recalled the Cambridge Analytica episode, which leaked personal data and helped to manipulate public opinion and help in the election of Donald Trump, in 2018, in the United States. “This case highlighted the importance of protecting user data, as it is often used to manipulate information and disseminate it quickly. This is the point: the spread of fake news gets so fast that we don't have time to moderate it and prevent its impacts”, she said.

“The most sensitive issue is that generative AI works with image, audio, videos, which are super convincing”, said Diogo Cortizo, recalling the episode in which an image generated by AI showed Pope Francis wearing a modern coat. “In some cases, like this one, even a specialized journalist from a fashion magazine was unable to distinguish and published the photo. Technology will evolve very quickly, and many scientists believe that we are entering a synthetic reality environment”, concluded the professor.