ChatGPT debate about AI in research during Hospital Week at Westmead Hospital

In a debate about the merit of generative artificial intelligence, or ‘AI’, in research; who better to argue for the affirmative side than ChatGPT itself?

This was the thinking of the affirmative team, consisting of Dr Poonam Balaji, Associate Professor Mark Douglas, and Dr Isabella Breukelaar, during the ‘Great Research Debate’ – the highlight of the Research Symposium’s events during Westmead Hospital’s Hospital Week.

Tasked with the statement ‘Generative artificial intelligence will change research for the better’, both sides took turns fiercely crafting their arguments and presenting them to a packed audience at Westmead Hospital in Western Sydney Local Health District (WSLHD).

ChatGPT seemed present in the very air of the auditorium during proceedings as attendees were left to question whether even the speaker’s biographies were penned by humans or AI-generated.

Debate swayed between the role of AI in enhancing efficiency by taking over menial tasks, to its responsibility for devolving the human race and robbing the next generation of formative education and learning skills. From the democratisation of information to the pervasion of racism and bias by automated systems. Even the role of Alexa in a modern remake of 2001: A Space Oddessy was called into question.

The affirmative team focused on the benefits of viewing AI as a 24/7 research assistant, who could carry the brunt of the basic research to free up more time for advanced work.

“These tools can process and analyse vast amount of data – which would not otherwise be humanly possible,” Dr Balaji said.

“By harnessing its ability to detect patterns and relationships within complex data, researchers can uncover hidden correlations and make new discoveries that can drive scientific progress.”

The negative team, consisting of Dr Caitlin Finney, Ms Parul Saini and Mr Heeva Bahariou, spoke of the impact of reliance on ‘inside the box’ thinking that stems from heavy use on tools like ChatGPT.

“I think the argument the enhanced data processing AI offers is missing the element of advanced question asking that is required,” Mr Bahariou said.

The only AI in medicine should stand for ‘actual intelligence’ because nothing beats the good old human touch – even if it means occasionally misreading chicken-scratch handwriting.

“When you learn how to build something from scratch, you feel that ownership of it, and that’s really important. You get this kind of immersion in the project that doesn’t happen if you’re detached from it [by using AI]– you don’t really want to build your dream house by just putting a roof on top of someone else’s work. That involved feeling means you want to go deeper into problems.”