What opportunities and risks does ChatGPT create for improving communication practices in research? The science communicators at Animate Your Science recently presented a practical article with a very optimistic perspective. We reached out to authors Dr Khatora Opperman and Dr Tullio Rossi to hear more.
Based in Adelaide, Australia, Animate Your Science offers animation services and SciComm training programs. With writers such as project manager Khatora Opperman and founder/CEO Tullio Rossi the blog continuously presents useful know-how, tips, and recommendations for their followers. In February, the article How to use Chat GPT: Opportunities and Risks for Researchers defined ChatGPT as the researchers’ “Personal Research Sidekick.”
Hi Khatora and Tullio! What features would you like to see in future updates of ChatGPT?
Khatora: “As a language model, ChatGPT is already much more powerful than before. However, I think several features could be added to make it even more useful for researchers. Specific training in the scientific text would be beneficial to improve its ability to understand and interpret scientific literature and technical jargon. Some have suggested features that improve ChatGPT’s ability to perform complex tasks like data analysis or literature review. However, I am skeptical about pushing language models in this direction. There is a fine line between augmenting human abilities and replacing them altogether!”
Tullio: “I think that the most significant limitation of ChatGPT today is that, although excellent at writing just about anything in a very confident way, it is not afraid to make stuff up. For example, I asked it to list the top papers in my Ph.D. field of research and it spits out a totally plausible list of article titles complete with authors and journal names but all of them were made up! Not a single one was real! It makes sense – it’s a language model, not a truth model. So, a feature I’d like to see in the future is the ability for Chat GPT to access the internet and reference its sources similarly to how Bing Chat does. Accuracy matters, and I am afraid that many people are not fact-checking the model’s outputs enough”.
How have your professional lives changed so far due to ChatGPT’s launch on 30 November 2022?
Khatora: “Conversational AI has been a game-changer for me personally. ChatGPT has become a useful day-to-day tool for brainstorming, accelerating my writing, and finding that word I just can’t think of! Whether it’s a meta description, an email, a social media post, or something else written, ChatGPT’s ability to suggest ideas, complete sentences, refine paragraphs, and offer synonyms has helped save me time and improve the quality of my writing. I also really like ChatGPT’s ability to adapt to my requested writing style, meaning I can focus on my ideas and creativity rather than on perfecting my writing. Overall, ChatGPT has become a valuable communication tool for me, and I’m excited to see how it will continue to evolve in the future”.
Are there any practical issues with ChatGPT that you find more problematic than others?
Tullio: “I’ve just returned from a business conference where somebody gave a presentation entitled “How to Use ChatGPT to Create 10 Blog Posts in Under 60 Mins”. My first reaction was, “WOW! I want to learn this method!”. Then it got me thinking…
Imagine how many people worldwide now use this tool to publish stacks of mediocre articles. It means that in a matter of months, the internet will be flooded with low-quality content with serious accuracy problems. How will search engines be able to differentiate AI-generated content from human one? I think that it will be a real problem, and it might make search engines way less reliable at providing accurate information.
After ruminating on this idea for a couple of days, I concluded that writing high-quality, long-form human-generated content has never been more important. This is what we will stick with as a company at Animate Your Science”.
How do you think ChatGPT will change the intellectual domain of international research in the long run – seen from a more philosophical point of view?
Khatora: “This is a difficult question, it’s hard to predict how the landscape will change in the long run. ChatGPT has the potential to revolutionize the way we approach to research; however, I fear that we may become excessively reliant on language models and AI more generally, leading to a loss of independent critical thinking and basic writing skills. Much like how Google Maps has made many people reliant on technology for navigation, or how to autocorrect has made many forget how to spell. AI can assist researchers in various tasks, but it should not replace the human element that is necessary for generating new ideas, designing experiments, and interpreting results. In the end, scientific research is ultimately a human endeavor and should remain that way.”