Artificial intelligence (AI) tools like ChatGPT and Copilot are now freely available to chemistry undergraduates and PhD students. From finding research papers and reactions to summarising concepts to writing reports and emails, such tools appear to save students’ time and make them more productive.
However, some researchers worry that relying on AI tools may stop students developing key skills, such as critical thinking and problem solving. A recent study found that nearly 30% of science, technology, engineering and maths (Stem) dissertations submitted in the US this year contained AI-generated text. This is a sharp increase from just under 2% in 2023 when ChatGPT became publicly available. Those behind the work noted that AI is ‘displacing a potentially important portion of the work through which scientific expertise is developed’. And earlier this year, the Higher Education Policy Institute (Hepi) – an independent education thinktank in the UK – released the results from a survey of over 1000 full-time undergraduate students on how they’re using generative AI (genAI) tools. The survey found that over 95% of students used AI in at least one way, which has increased from 92% and 66% in 2025 and 2024, respectively. Additionally, 94% of students said that they used genAI to help with assessments – up from 53% in 2024
What is AI?
Artificial intelligence (AI) is an umbrella term often incorrectly used to encompass a variety of connected but simpler processes.
AI is the ability of machines and computer programmes to perform tasks that typically only humans could do, such as reasoning, responding to feedback and decision making.
Generative AI is a newer variant of AI that analyses and detects patterns in training datasets to generate original text, images and videos in response to requests from users. ChatGPT, Microsoft Copilot, Google Gemini and X’s Grok are all examples of chatbots that use generative AI.
Large language models or LLMs are a type of deep learning program trained on large amounts of data to understand and generate language. LLMs learn patterns in text by predicting the next word in the sequence and these models are now able to write prose, analyse text from the internet and hold dialogues with users.
Neural networks are an interconnected array of artificial neurons, akin to biological brains, that identify, analyse and learn from statistical patterns in data.
Machine learning is a subset of AI that allows machines to learn from datasets and make predictions based on new data, without programmers explicitly asking it to do so. Machine learning models improve their performance as they receive more data.
Deep learning is an enhanced type of machine learning that uses neural networks with many layers to analyse complex data from very large datasets. Applications of deep learning include speech recognition, image generation and translation.
Chemistry World spoke to several students about how genAI has been changing learning and how and why they’re using AI tools.
Putting it to the students
Dion Josh, a second-year chemistry undergraduate student at the University of Oxford, UK, says that he mostly uses AI agents to break down concepts so that they are easier to understand. He says that this works well for inorganic chemistry, which often relies on learning trends and facts, but notes that using AI tools to figure out organic reaction mechanisms is trickier. To overcome this, Josh often resorts to asking several AI agents the same question. ‘I [also] ask it to show me it’s working and ask it to reason through why the reaction is taking place,’ he says.
It’s only the really intense questions that I don’t know – that AI can’t teach me – that I need the professors
Josh highlights how using AI tools during his degree has also made his tutorials more useful. ‘I don’t have to ask the little questions that I don’t need a professor for,’ he says. ‘It’s only the really intense questions that I don’t know – that AI can’t teach me – that I need the professors.’
Trawling through and summarising recent research literature is ripe for using AI tools. However, Ellie Yates, a PhD student at the University of Sheffield, UK, working at the intersection of materials chemistry and plasma medicine, notes that she often finds using these tools for such tasks unhelpful. ‘While it can be good for a quick search to get a list of topics to research yourself, it does more often than not produce lots of “hallucinated” answers which could lead your research in the completely wrong direction if you’re not careful,’ she says. Yet, Yates recognises that ‘AI agents are improving rapidly’. ‘It’s essential for students and researchers to remain AI literate, understanding both the capabilities and the limitations of these tools rather than just ignoring or avoiding them.’
It’s essential for students and researchers to remain AI literate, understanding both the capabilities and the limitations of these tools rather than just ignoring or avoiding them
Yates also worries that students are using AI tools to help with written assessments like lab reports. ‘Students need to be very careful that they don’t miss out on the opportunity to refine their writing skills during their degree,’ she says. ‘I completely understand that writing lab reports can be very time consuming and difficult to do [but] it is a very important skill which, especially when it comes to how you analyse data, shapes how you think as a scientist.’
Yates says that she tries to use AI tools ‘relatively sparingly to avoid becoming dependent on [them]’.
Meanwhile, Jasper Butcher, who has just finished his PhD on protein design in Nobel laureate David Baker’s lab at the University of Washington, US, says he uses AI tools ‘for everything’. He primarily uses Anthropic’s Claude to generate code and run experiments as part of his research. ‘I would say there’s a really positive angle to some of this work where it’s helping us do good science at a faster rate,’ he says.
I had so much fun building [problem sheets for students] because all of a sudden I could put together more creative problems for people to solve with much less effort
AI tools have also helped Butcher to develop problem sheets for his students. ‘I had so much fun building them because all of a sudden I could put together more creative problems for people to solve with much less effort,’ he says.
Reflecting on his time as an undergraduate, Butcher thinks that AI tools will have changed the way that students now collaborate. For him, working with his peers on assignments ‘made the whole experience shared’ and that it was ‘so satisfying to get through [assignments] together’. ‘You couldn’t really do [a degree] alone in some ways,’ he adds.
He feels that students may now be ‘one-shotting’ their degrees – working in isolation with an AI agent, rather than with other students. ‘That’s a real shame,’ he says.
‘I’m personally very fortunate to be in a situation where I have had to [as part of my studies] sit with really difficult problems,’ he says, adding that the skills needed to do that now ‘feel harder to synthesise’.
Teaching in the time of AI
Fraser Scott, who researches chemistry education at the University of Strathclyde in the UK, notes AI use is ‘rife throughout undergraduate all the way up to PhD student’. ‘[AI] has emerged very quickly – too quickly for educators and education institutions to respond to adequately,’ he says. ‘I think the response [so far] has been focused more on safeguarding assessments rather than understanding or promoting genAI as an education tool.’ In the Hepi survey, less than half of students (48%) felt that they were being taught how to use AI effectively, despite more than 68% agreeing that these skills are ‘essential’ to ‘thrive in today’s world’.
‘As educators, just as we’ve got a responsibility to teach our students how to use a textbook appropriately or a calculator appropriately, we need to teach them how to use genAI appropriately,’ says Scott.
He also points out that educators often have strong opinions about students using AI tools ‘but I bet every single one of them uses it very regularly themselves’. ‘There’s this sort of dichotomy of use,’ he says.
‘It’s just how much they’re leaning on AI, rather than are they using AI or not,’ Josh says. ‘All the undergrads that I’ve spoken to don’t understand how anyone’s got through university without AI because that’s what we entirely rely on.’
