Will AI take over grant writing?

Now that anyone can drop a draft into a chatbot and get back cleaner text in seconds, what is left for a grant writer to do?

It's a fair question, we have seen the meteoric rise of these tools in the last couple of years. And I want to be totally transparent, I do use these tools and NO I don’t think they are coming for my job.

Let me start by agreeing with the part that is true. If you have written a proposal and you want the English tidied up, the sentences smoothed out, the paragraphs tightened, a good AI tool will do that quickly and cheaply. You should use it. I use it too. There is no point pretending otherwise, and any grant writer who tells you their value is turning rough text into neat text is selling you something you can now get for free. I actually think this is great, especially for applicants whose first language is not English, as this provides a more level playing field.

But that is not really what a grant proposal needs, and it is not what I am for.

Here is the thing about a strong application. The problem is almost never the grammar. The problem is the argument. A proposal gets rejected because the aims are too ambitious for the timeline, or because the most exciting result is buried on page nine, or because the applicant assumed the reviewer would understand something that actually needs three sentences of explanation. It gets rejected because the case for why this work, by this person, right now, was never quite made.

None of those are writing problems in the way a chatbot understands writing. They are judgment problems. And judgment is the thing I am actually paid for.

When I read a draft, I am not mainly reading for typos. I am reading as a reviewer would. I am asking where this proposal will lose a reader who is tired, sceptical, and working through a stack of thirty applications. I am asking whether the ambition is pitched right, not too timid, not so grand it stops being believable. I am asking whether the science holds together, which I can only do because I have a research background and have sat on the other side of the table. I am asking what this particular funder rewards, because an ERC panel and a national funder are not looking for the same things at all.

A language model can make your text read more smoothly. What it cannot do is tell you that smoother is not the same as more fundable. At the level most of my clients are working, that gap is crucial. Two proposals can be equally well written and one gets the money. The difference is rarely the prose, it is the thinking underneath it.

When you hand your proposal to me, you are handing it to a person who will still be here after you submit. If it does not get funded, I am the one who talks it through with you and helps you try again. If a reviewer raises something odd, I have an opinion, because I have seen it before. That relationship, someone who understands your work and is accountable for the outcome, is not something you get from a text box. It is a large part of what you are actually buying, even if it does not show up on the invoice. We can have a conversation, we can ‘debate’ your work, I can push you and encourage you and listen, in a way that only a human can.

So no, I do not think AI is going to take this job. But I do think it changes it, and I would rather be honest about how.

The part of grant writing that was only ever about producing tidy words is genuinely shrinking. If that is all someone offers, they should be worried. The part that is about judgment, strategy, understanding the science, and standing behind the result is not shrinking at all. If anything it is becoming more valuable, because as the easy layer gets automated, the thing that is left is the thing that was always hardest to find: someone who knows what a fundable proposal actually looks like and can tell you honestly where yours falls short.

I use these tools every day, and they make me faster. They handle the parts that used to eat my time, which means I can spend more of it on the parts that need a human who has done this before. That feels like the right way round to me. The tool does the tidying. I do the thinking.

So if you are wondering whether it is still worth working with a grant writer now that the machines can write, here is how I would put it. Run your draft through whatever tools you like. Then bring it to someone who can tell you the things the tools never will: whether your argument is strong enough, whether a reviewer will believe you, and what you need to change before you hit submit.

That last mile is where grants are won and lost. And it is still, very much, a human job.