AI is shaping the very nature of academic disciplines

By Hugh Halsey

Everyday there are new articles and opinions on the impact AI is having on our everyday lives. It’s impossible to escape. Very few would deny the impact AI is having (both good and bad) but AI is having impacts beyond what we can see.

A recent study revealed that AI is even changing the way we speak. Words that previously had limited use in conversation such as ‘delve’ and ‘boast’ have seen huge spikes in use since the introduction of AI in 2022. Why? Because of the dialects that ChatGPT was trained on use these specific words more than the average person in unstructured conversation. We are literally speaking more like ChatGPT.

However, away from the pressures of the consumer market, AI has a different story to tell. AI has proved to have many valuable applications in scientific research. AI brings faster analysis on ever larger datasets, revolutionising ecological research with new insights into the biodiversity crisis and climate change. AI has also proven valuable in medicine, being able to detect cancer tumours earlier than human doctors. Even my own masters project in Biology this year is using AI to create new protein structures with medical applications. 

The changes that science is seeing are huge and it doesn’t show signs of slowing down, with new applications being introduced all the time. I read in the FT recently an opinion piece laying out a case for AI in Europe. An area that we are reminded frequently the EU is behind in compared to the US. When the AI bubble in the US bursts, Europe might prove well positioned to build more trustworthy AI models to be used in science and engineering. AI technology that has a specific purpose rather than to serve a consumer market.

But not all are excited about the changes AI could bring to science. Many ecologists for example are concerned about how AI is changing what research is being conducted. Fieldwork is increasingly seeing a decrease in published papers and available funding. AI brings so much computational power in analysis that it incentivises scientists to publish papers analysing current data rather than collecting new data. Ironically, many see ecology as increasingly moving indoors with a disconnect from the natural environment which it studies. 

There is immense pressure to use a technology that may not be helping science to answer the most important questions

A recent article from Nature suggested that whilst AI is benefiting individual scientists in their career with increased computing power to spot trends, it incentivises specific areas of research; analytical research is more rewarded than data collection. This does not benefit science as a whole. Ecologist, Bill Sutherland from the University of Cambridge put it very well when speaking with Nature. “Supposing you do your PhD and you spend all your time doing fieldwork, and the person sitting next to you has been extracting data. After three years, they might have published in increasingly highly ranked journals, while you’re still in the Amazon catching fish”. 

What will that do to the future of research if AI changes which areas of research are rewarded in journals? Another recent study showed that citations of papers that used AI were far higher compared to those that didn’t. For a scientist to increase their chance of promotion and securing future funding, there is immense pressure to use a technology that may not be helping science to answer the most important questions.

The Royal Society also has much to say on the matter as well. Laying out the challenges with AI data being hard to reproduce due to the nature of the technology. It is impossible to know quite what the technology has done to draw its conclusions. This goes against the very bedrock of science; conclusions and results are reproducible. It gives the research credibility.

AI is obviously not going anywhere and promises many great benefits to scientific research. Many might argue that this is where AI can have the largest positive impact for society. But as we continue to discover what this technology can do for science, it may change the direction of fields of research in ways beyond what we intended.

Image: GlacierNPS via Wikimedia Commons