ChatGPT: five priorities for research | Nature Comment

ChatGPT: five priorities for research

Conversational AI is a game-changer for science. Here’s how to respond.

ChatGPT is a large language model (LLM), a machine-learning system that autonomously learns from data and can produce sophisticated and seemingly intelligent writing after training on a massive data set of text. … This technology has far-reaching consequences for science and society. Researchers and others have already used ChatGPT and other large language models to write essays and talks, summarize literature, draft and improve papers, as well as identify research gaps and write computer code, including statistical analyses. Soon this technology will evolve to the point that it can design experiments, write and complete manuscripts, conduct peer review and support editorial decisions to accept or reject manuscripts.

Well, I’m not so sure that ChatGPT will be automating all of my work soon, but the article is thought-provoking. I feel a bit behind here as I haven’t yet personally used ChatGPT in my research, but it just told me why I should care about LLMs:

Large language models like ChatGPT have the potential to significantly impact scientific research by enabling new approaches to data analysis and hypothesis generation. For example, models can be trained on vast amounts of scientific literature and other data sources, allowing them to suggest new hypotheses and connections based on patterns in the data. They can also be used to automate certain tasks, such as data preprocessing or experimental design, freeing up researchers’ time for more creative and strategic work. Additionally, language models can facilitate communication and collaboration among researchers, enabling them to more easily exchange ideas and knowledge across disciplines. However, it’s important to note that the use of such models in scientific research should be carefully considered and evaluated, as they may also introduce biases and limit interpretability in some applications.

The above is ChatGPT response to my prompt: ‘How will large language models like ChatGPT change how scientific research is performed.’ (I’m quite happy with that, and, if nothing else, will probably use it to write some content for the next IGDORE newsletter :smiley:) If you haven’t already done so, try out ChatGPT yourself at:

The original article calls for ‘organiz[ing] an urgent and wide-ranging debate’ about the disruptive potential of LLMs, and suggests discussing the following issues:

• Which research tasks should or should not be outsourced to large language models (LLMs)?

• Which academic skills and characteristics remain essential to researchers?

• What steps in an AI-assisted research process require human verification?

• How should research integrity and other policies be changed to address LLMs?

• How should LLMs be incorporated into the education and training of researchers?

• How can researchers and funders aid the development of independent open-source LLMs and ensure the models represent scientific knowledge accurately?

• What quality standards should be expected of LLMs (for example, transparency, accuracy, bias and source crediting) and which stakeholders are responsible for the standards as well as the LLMs?

• How can researchers ensure that LLMs promote equity in research, and avoid risks of widening inequities?

• How should LLMs be used to enhance principles of open science?

• What legal implications do LLMs have for scientific practice (for example, laws and regulations related to patents, copyright and ownership)?

Do you have any comments about the above issues? Or other thoughts about the use of LLMs in research? Have you already used ChatGPT in your research? If so, then do share your thoughts and experiences here (I’ll add some of my own soon)

Relatedly, the Night Science podcast has a good episode that discusses the use of AI in medicine:

1 Like

Some perspectives on the use of LLM’s in university teaching and assessment.

Large language models challenge the future of higher education

One question that arises is whether and how higher education should react. Should universities ban its [ChatGPT] use? Or should academics instead accept that language models will become integral to their professional toolkit, and incorporate them in our teaching and assessment practices?