Are we too reliant on AI?
A new study by the University of Surrey is investigating over-reliance on AI has received a grant of £260,000
A new study on how generative artificial intelligence (GenAI) is impacting the education and work of healthcare researchers has received a £260,000 boost from UK Research and Innovation (UKRI), as part of a £4 million research funding programme investigating how AI is impacting science.
Led by the University of Surrey, the study will be exploring how healthcare researchers at all levels are using AI, highlighting the benefits and potential risks they may face.
The study will test research students’ reliance on AI by tasking them to analyse data that deliberately includes hidden errors.
By observing whether they can spot these mistakes, the study aims to determine whether GenAI ultimately supports or weakens research integrity.
AI has become a ubiquitous tool that can provide quick answers in areas where specialist expertise is usually required.
However Dr Maupin says that AI's confidence can lead to critical mistakes
"These tools like to give quite confident answers and if you use the scenarios you're not really an expert in, you might kind of miss some things and make consequential mistakes."
He describes two different types of people when it comes to over using AI for healthcare studies.
"There's kind of two ways that we're seeing artificial intelligence be used in research, one is the bad actors who are using it to kind of falsify or just generate studies ad hoc and just type in a prompt to ChatGPT and say to write a study about this.
The other is probably the well-intentioned people who are just using it as a tool and they might, you know, end up getting carried away or using it in a manner that just leads to mistakes because they might not have a full understanding of what it is that they're doing."
Dr Maupin warns that over usage of AI could undermine the reliance of healthcare studies as a whole.
"I think it just undermines the trust in the the research output. So you can have studies that might find a significant association between some medication, some disease, but it was a poorly done study that hasn't accounted for all the factors
We can see there's a big thing going on in the US around Tylenol and autism at the moment. It's a very complex area.
The last thing you want to do is have someone without that significant understanding trying to conduct research in this and flooding the environment with this low quality research that might have a significant result and end up kind of causing some fear or anxiety and leading people down the wrong path."