Last week, KFA's award-winning senior editor Chelsea Adams wrote a great op-ed on using AI to write grants, ARTIFICIAL INTELLIGENCE AND GRANTS. WILL IT WORK? There's just one problem. I totally disagree.
I follow a fairly well-known science fiction author on social media. Lately, he’s been playing around with having AI write stunning reviews of his work in the voice of other authors. The experiment is tongue-in-cheek, but the results have been fascinating. And, it’s important to pay attention to sci-fi writers. They see the future in ways most of us don’t. I’m more of a fantasy nerd, but I have a deep appreciation for the prescience of sci-fi.
In addition to grants, I also occasionally write 300+ page technical reports for government contractors (quite the opposite of both sci-fi and fantasy, the driest sort of reading there is). A few years ago, I was involved in documenting a project that extensively used AI to analyze narratives in a massive, national database. I saw right then and there that it wouldn't be long before AI would be writing grants. (I'm afraid AI could also be scoring grants, perhaps, before AI is writing them. But that is a problem and moral dilemma for another day.)
Grant writing is formulaic
If you've taken our course, you know that grant writing can be very formulaic, if you know the formula. If you don't know the formula, take our course.
I can easily see how, using our formula, a program like ChatGPT could generate a lot of grant narrative in a short amount of time, cutting a lot of time-consuming work out of the grant writing process. The tricky part is figuring out how to train the AI to follow the formula and doing that will require a high-level understanding of what goes into the formula in the first place. Again, you can gain that knowledge in our course.
Let’s revisit what ChatGPT had to say when Chelsea asked it, “Can AI be used to write grants?”
“Yes, AI can be used to write grants. AI algorithms can be trained on examples of successful grant proposals, and then be used to generate new proposals based on the input they receive. However, it is important to note that AI is not capable of understanding the context and nuances of grant writing, and the proposals generated may still require significant editing and revision by a human grant writer.”
I believe every word ChatGPT said is true. And I’ll break it down, step-by-step.
AI and the grant-writing process
First, AI has not yet evolved the ability or creativity to design projects. Frankly, a lot of people haven’t developed this talent either and fortunately that’s what makes us worth our weight in gold. We can start with a blank page and design an award-winning project. We also teach this talent in our handy Quick Start Guide to Logic Model Planning. AI’s inability to design projects is what requires AI to be trained from an existing library of successful proposals. In reality, most grant writers use this approach themselves. But there are some legal and ethical issues there.
Novice grant writers often try to borrow project designs from funded proposals. (STOP. Before you get any bright ideas about firing off a bunch of FOIA requests, keep reading.) Grant writers have to take care not to violate copyrights or outright steal others’ ideas. Expert grant writers like us have our own extensive library of work to pull from. We write a few major anchor proposals for a client and repackage that content over a period of time to gain extraordinary efficiencies and garner big bucks for our clients. There lies the major hurdle to feeding AI successful proposals. To do it, you have to have your own source material. Once you have it, you can enjoy the same extraordinary efficiencies we do. But getting there is going to be more than half the battle.
Project design is just the first hurdle. ChatGPT admits that AI is not capable of understanding the context and nuances of grant writing. AI is inquiry based. In order to get AI to work, you not only have to feed it the right source material to work with, but you also have to ask it the right questions. What ChatGPT is really saying is that it can’t read an RFP and write a responsive proposal. If a grant writer has the right source material to train AI and knows how to process an RFP (a major skill acquired in our course), a grant writer can ask AI the right questions and steer this workhorse in the right direction to generate narrative.
Let’s assume for a moment that we have a massive content library to work with and expert RFP processing skills. What is to stop us from using AI then? Well, ChatGPT says, “the proposals generated may still require significant editing and revision by a human grant writer.” That is because while grant writing does lend itself to a formulaic approach, grants still must be compelling and persuasive. While most of our course focuses on how to reduce grant writing down to a relatively simple formula, we still teach an entire section on how to write in a compelling and persuasive way. We use credibility building, putting data in context, and leveraging analogies that evoke an emotional response to do that. AI is still several generations away from evolving that level of sophistication. But I do believe it will get there eventually.
And what about budgets?! Oh, please. Expert budgeters have been using Excel formulas to take the work out of budgeting for decades. AI’s older, clunkier cousin will teach it to budget grants in no time. In fact, you can bet this is precisely where I will start using AI to increase my grant production efficiency.
Are you ready for AI?
Wait! What? Yes, I have every intention of getting on the AI bandwagon. I will admit that I’m going to start by having my teenager teach me to use it. But I know I possess the expertise to put it to work for us and our clients, because I understand its limitations, I have the right source content to train it, and I’m going to ask it the right questions. I recommend joining our course to gain these skills for yourself before you get left in the dust.
Now, excuse me while I go talk Chelsea down off the roof.