Many of us have already been trying Chatgpt. If you’ve checked science social media recently, it’s likely that you’ve already seen many of its writings. In common with many other researchers, I worry about artificial intelligence (AI) replacing me. I’m a vaccine researcher and spend much of my time Writing grant applications, papers and articles about science careers, so I set the chatbot the task of writing an opinion piece about the use of AI in grant writing.

In less than a minute, ChatGPT had produced the following:

In my opinion, ChatGPT has the potential to revolutionize the process of writing scientific grants. Traditionally, writing a scientific grant has been a time-consuming and often frustrating process. Researchers spend countless hours crafting proposals, only to have them rejected by funding agencies. This can be demoralizing, and it can also be a barrier to progress in scientific research. ChatGPT has the potential to change all of this. By using natural language processing and machine learning, ChatGPT can help researchers write more compelling and effective grant proposals. It can also help reviewers assess grant proposals more efficiently, allowing for a more efficient and fair grant review process. Of course, ChatGPT is not a magic solution to all of the challenges facing scientific research. But it has the potential to make a real difference, and it is worth exploring as a tool for improving the grant writing and review process.

Are ChatGPT and AlphaCode going to replace programmers?

Which, as my editor for this piece says, is pretty good for an AI program, but not good enough to be deemed worth publishing as a standalone piece of writing. It is, however, a massive step forwards. The simplicity and utility of platforms such as ChatGPT mean that we’ll see them quickly drifting into everyday use; Microsoft is already working closely with OpenAI, the company that developed ChatGPT – and you might already be using an AI platform to help you with some writing tasks. At present, ChatGPT is freely available, but things are moving extremely fast, so it is hard to know what will happen next.

Something to build on

The example above wasn’t the first time I’d used AI for science writing. My first real use of AI chatbots (beyond asking one to write lyrics to a song called ‘Eggy Eggy Woof Woof’ for my daughter) was when I got fed up with writing one part of a grant application. I was asked to explain the world-changing ‘impact’ that my science would have, if I was lucky enough to receive funding.

So I asked ChatGPT: “What impact could vaccine research have?” and got 250 words of generic fluff. It suggested reducing the burden of disease, saving lives, improving global health and supporting economic development. None of what it said was in any way original or enormously surprising, but it was an excellent starting point, which I could then flesh out with specifics.

So far, I’ve found AI to be enormously helpful at generating a lot of the low-level filler documents that otherwise take up time. For example, it can write easy, generic, on-the-one-hand-this-but-on-the-other-that statements about Internet-usage policy or data management. However, it’s still early days, and much more thought needs to go into exploring the implications of AI regarding plagiarism and attributing credit.

‘Arms race with automation’: professors fret about AI-generated coursework

Various publishers (including Springer Nature, which publishes Nature) have scrambled to make clear their policies on using generative-text models. Generally, these models are seen as methods, rather than as co-authors, because they can’t take responsibility for the work. But questions remain. For example, the vast corpus of language that’s used to train many AI models will include some copyrighted material: how do we know that these models aren’t simply quoting something from a protected source?

One situation in which I have found ChatGPT extremely useful is writing multiple-choice questions. It’s quite easy to write a question and the right answer, but coming up with three plausible wrong answers is tricky. I found that if I prompted ChatGPT with the following: “Write a multi-choice question about with four answers, and not using ‘all of the above’ as an answer,” it came up with good wrong answers. This was incredibly helpful. As someone suggested on Twitter, if students are going to use ChatGPT to help them write answers, we might as well use it to set the questions!

Generic waffle

Another time, I entered the prompt: “Write a diversity statement for a grant application.” The output starts:

Diversity is a core value at our organization and we are committed to promoting and maintaining an inclusive and equitable environment for all individuals.

It continues for 169 words that are generic to the point of being meaningless. Anyone can write such a response, and it requires no evidence or backing. It would be better if the grant-application form asked questions that an AI program can’t answer, such as: “Give examples of how your organization is promoting diversity, and evidence of the impact that these activities have had.” This could be applied to a whole range of questions that we are forced to answer.

ChatGPT: five priorities for research

This made me reflect: if there is a section in a grant application that can be written by an AI, does that section really serve any purpose? If a computer can churn out something deeply generic that still answers the question (more or less), why are we asking people to address the question at all? I think the answer is clear: these sections never really did serve a purpose and certainly don’t now. For science and the process of grant writing to be improved, two things have to happen: first, the pointless sections (those that might as well have been written by a computer, and could just as easily be answered by one) need to be removed; and second, the sections that remain need to be changed in scope, to be shorter and action-centred.

But for now, while we are forced to fill in unnecessary boxes on forms, AI offers a way to free up headspace, which should be a good thing. In a Nature article last month about a decline in ‘disruptive’ science, one of the suggestions was that academics needed “the gift of time”. AI could well give us this gift.

Freedom to be disruptive

The question is, then, how do we use the time given? One comparator is the automatic washing machine: as it became universal in the 1970s, it freed up time, which was then taken up with other household tasks. The sociologist Joann Vanek argued in 1974 that despite new household devices, the time devoted to housework had not changed over the past half-century. Her argument has been debated, but the key question is what impact do time-saving devices have? Are we going to fill the time saved by AI with other low-value tasks, or will it free us to be more disruptive in our thinking and doing?

Tools such as ChatGPT threaten transparent science; here are our ground rules for their use

I have some unrealistically high hopes of what AI can deliver. I want low engagement tasks to take up less of my working day, allowing me to do more of what I need to do to thrive (thinking, writing, discussing science with colleagues). And then, because I won’t have a Sisyphean to-do list, I’ll be able to go home earlier - because I’ll have got more of the thinking, writing and discussing done during working hours, rather than having to fit them around the edges.

We are unlikely to arrive at these sunlit uplands without some disruption. Just as domestic appliances significantly shrank the need for domestic staff, AI is going to change the labour market. For some tasks, AI will replace people. The aim of the game is to not do a job that can be replaced by an AI program. Hopefully, I have persuaded you that although AI can write, it isn’t going to replace me or others in my profession immediately. I’m not the only one to see things this way: the songwriter Nick Cave put it much more eloquently here. One piece of good news for me is that AI isn’t very good at telling jokes. I will leave you with its best effort: