Trust, but Verify - A Petrus Development Show Episode on AI in Nonprofit Fundraising

What if you could have an intern who works 24/7, never complains about tedious tasks, and costs almost nothing? While finding human help for repetitive fundraising tasks can be challenging and expensive, AI tools like ChatGPT are revolutionizing how nonprofit development offices operate.

Join Andrew and Rhen as they explore the practical ways artificial intelligence can become your most valuable (and affordable) team member. They'll share specific strategies for leveraging AI in fundraising while highlighting the crucial boundaries every development professional needs to understand.

Show Notes:

In this epsiode, Andrew and Rhen share concrete ways that successfully use AI in their fundraising lives.  They offer specifics about which platforms are best for which tasks, and they suggest ways fundraisers can take advantage of AI in a wide range of tasks including writing, data analysis, graphics, and more.  

Andrew highlights examples of times when AI is not reliable or effecitve.  AI can never replace the fundraising success of human interaction and relationship building, and AI is not infallible.  Andrew encourages fundraisers to embrace AI's capabilities while maintaining healthy skepticism about its limitations.   

INTERVIEW TRANSCRIPT

05:08.01 Host Well, howdy, everybody. Welcome back to the Petrus Development Show. I am Rhen Hoehn from Petrus Development. Joining me today, big man, the president and owner of Petrus, Andrew Robison. How's it going, Andrew?

05:19.83 AROB It's going great. How about yourself, Rhen?

05:21.62 Host I am doing very well. I am enjoying your American flag fish hat, celebrating Memorial Day. All right.

05:28.54 AROB There you go. God bless America.

05:30.74 Host Love it.

05:33.30 Host We didn't talk about an opener for this.

05:35.11 AROB No.

05:36.71 Host Anything you want to cover specifically?

05:39.38 AROB And...

05:47.17 AROB I have no idea.

05:50.14 Host Trying to think if I've done anything interesting with chat.

06:05.12 AROB I've written a couple of— oh, you got something?

06:05.23 Host Okay.

06:07.09 AROB All right.

06:07.36 Host Um, I think I can go with it. Did you have a good Memorial Day weekend in your household?

06:14.13 AROB I reckon it was. I'm getting ready for this. Um, I'm doing— I'm leading a retreat to conference this week, um, for campus ministries in the state of Texas.

06:25.72 AROB So we're going to be talking about the campus ministry playbook, which is a fun project. If you want to learn more, if you're a campus minister— campus minister— go to campusplaybook.com. You can download your free copy, but I'm going to be leading a group through that, um, kind of a workshop through the playbook and doing about that. So yeah, we did some fun stuff for Memorial Day, hung out with the kids, swam in the pool.

06:48.85 AROB Um, but nothing big and crazy because I got this, uh, retreat coming up.

06:53.48 Host And so we often travel from Memorial Day... boom... uh, go see some relatives, but they all came up to visit us earlier this spring. So we stayed home.

07:01.18 AROB Nice.

07:01.93 Host We had a chill, chill time, but my kids got really into something you probably haven't heard much about in the news. They love AI now for two specific, two specific purposes that they use it for.

07:12.38 AROB Oh, tell me more about this AI. Okay.

07:17.68 Host One is finding out the weirdest facts from history.

07:20.44 AROB Okay, good.

07:20.72 Host That's been a big thing they want to know about. We learned a lot about the emu war in

07:24.39 AROB Okay.

07:24.83 Host Australia, where they tried to kill all the emus and lost.

07:29.51 AROB Did the emus defeat them or did other people defeat them?

07:32.60 Host Uh, they failed to defeat the emus.

07:35.44 AROB Wow.

07:36.59 Host Because they were just hard targets. It was a whole thing.

07:39.01 AROB Okay.

07:39.56 Host We did this big deep dive on that. The other thing they've learned is they can get a lot of tips from Minecraft from ChatGPT.

07:46.61 AROB Ah, yeah.

07:47.62 Host So we did a deep dive on all of the different things you can do that are hard to find in Minecraft that they are now experts on.

07:55.98 AROB We've done that. My daughter Nora has— she plays this game called Pokémon Violet on the Switch.

08:01.14 Host Okay.

08:02.02 AROB And it's a pretty straightforward game, but there's definitely some quests that are built in. And so she's had me look up, "Dad, how do you find this thing? Or how do you unlock this Pokémon?" And so I asked chat and it tells me. I was like, "You want to figure that out on your own, right, kiddo?" She's like, "Never in a million years, Dad."

08:22.98 Host Nope.

08:23.19 AROB Okay, good. Yeah.

08:24.55 Host It's amazing what you can accomplish for the betterment of humanity with AI, I guess.

08:26.43 AROB Yeah. Yeah. Yeah. We do...

08:30.21 Host Oh my goodness. But that got me thinking, um, you know, if we can use it for fun things like that— generating coloring pages has been another favorite thing. You can, you know, generate a coloring page of your own town.

08:42.41 AROB Yeah, we do a lot of that.

08:42.77 Host That type of thing. Yep. Um, what can we use AI for in ministry? And I think there are certain things we can use it for and certain things we should definitely be aware of maybe and either not use it for or be very careful how we use it in that use case.

08:57.53 AROB Yeah. Mm-hmm. Yep.

08:58.96 Host So should talk a little bit about that today. We have a list of a few things like I said— when to use it, when to maybe avoid it or be very careful with AI like ChatGPT, like Grok is the X one.

09:12.59 Host Um, there are all kinds of them out there. It's very difficult to keep up with them at this point. Uh, but we'll mostly talk ChatGPT, I think, in this case, because it's the most common, most known one. Uh, so let's talk a little bit about when might you use AI in your nonprofit.

09:27.34 AROB Good. So from a fundraising standpoint, which I presume that's where you want to go with this, right?

09:31.55 Host Uh, exactly.

09:31.94 AROB But yeah, okay, good.

09:32.35 Host Yep.

09:33.21 AROB Uh, so it's a little bit— I think of chat or I think of AI as almost like my really low-cost intern, right?

09:44.44 AROB So there are multiple things in fundraising that, you know, in running a shop and, um, you know, in organizing, you know, your fundraising strategy that

09:44.65 Host Right.

09:58.29 AROB they need to be done. But gosh, if you had an intern who was really capable and really quick and didn't take coffee breaks and all of that stuff, that you'd be more than happy to hand these things off to them to do, right?

10:10.26 Host Right.

10:10.36 AROB And so I think of chat as my intern. Really low cost, $20 a month. I pay for the pro. Um, and I use it and I use a lot of the GPTs. A lot of people don't realize that there's, you know, chat, which is the kind of standard one.

10:24.96 AROB But then there's a lot that are, um, that you can use that are used for specific uses. What that does— what that means is they've just sort of programmed the model to respond in different ways.

10:36.55 AROB Um, and so I use, you know, one for— and in my work with Petrus, you know, when we bring on a new client, we have contracts that we sign and, um, things that we're negotiating, you know, with proposals and things like that. And so I'll use it to like kind of change the wording in a clause, you know, for a particular client and need things like that.

10:54.36 AROB So anyways... kind of the starting line is the base sort of level of use that I'll consider it for is as an intern, right?

11:06.18 Host And...

11:06.46 AROB So I think we're going to talk about like what some of those ways are, but the implication there is that we're not going to use it for certain things that I feel like I would need to, or the fundraiser or the executive director needs to do themselves, which we'll talk about in sort of the second half, right?

11:22.64 Host Exactly. And I think you can go very deep with this. You can get very sophisticated with this. I think let's just stick to some of the use cases that are approachable for somebody who hasn't done a whole lot with AI at this point.

11:33.57 Host And know that things change very fast. If we had done this episode a year ago, it would have looked pretty different.

11:39.86 AROB Yeah. Yeah.

11:40.03 Host Uh, some of the things that were, you know, just— things have changed.

11:41.63 AROB Yeah.

11:44.15 Host Chat has improved its answers. It's improved its memory. So you can say, "Hey, going forward, don't do this." And then it'll remember that. Or "I like this tone" or whatever the thing is.

11:54.95 Host Uh, you can tell that it's going to remember that for future queries, which is super nice.

11:58.81 AROB Yeah. Okay, so let's start with the very basic one with writing. Um, somebody told me one time that AI is to writing what the calculator was to math. So it's not that you turn over your writing to the robot— it's that you use the robot to help refine, help to brainstorm, help to reword your writing that you are doing.

12:24.11 AROB Just like you would use a calculator, right?

12:24.87 Host Yeah.

12:26.57 AROB I don't look at the calculator and it does the calculations. I have to be the one that knows what numbers I'm trying to find, that knows the formula. I put it together and then the calculator calculates the math much quicker than I could.

12:41.26 AROB AI chat can do the writing or the editing or the ideation quicker than I can, but it still requires me to direct it. AI is really good at mimicking.

12:52.14 AROB So it can mimic things that already exist and that it has found and stored in its memory— whether that's from you or whether that's from the internet, whether that's from other people using AI. AI isn't a creator— it's a mimicker. And so you think of it in that way and you can say, "I want to..." I wrote this. Uh, so for example, thank you letters, um, thank you emails, uh, appeal letters, newsletter articles, grant— right— grant applications, um...

12:52.27 Host Yeah.

13:21.53 Host Thank you.

13:22.60 AROB Uh, a lot of thank you letters, a lot of writing tools or writing that you need to do. You can use AI to brainstorm. "Hey, I want to write an appeal letter that, um, incorporates, you know, this particular message or this particular scripture as the theme.

13:41.09 AROB Can you give me some ideas for how this appeal letter could be written?" And it does 400 words in three and a half seconds.

13:52.51 AROB And then you take that, you read it and you say, "I like this stuff. I don't like this. I'm going to reword this." And then all of a sudden you've got an appeal letter that instead of taking you a day of sitting there coming up with it and brainstorming and all of this, you've got it done much quicker, but you still have your personal tone and input on it.

14:10.88 Host Exactly. And there are things to be aware of when it writes for you, right? It definitely has tendencies that repeat a lot.

14:16.90 AROB Yeah.

14:17.10 Host So I think what's kind of become famous is ChatGPT uses what's called an em dash a lot, which is like a long—

14:21.89 AROB Why it insists on using this in every sentence.

14:24.51 Host Every sentence. And another thing I noticed is it loves the kind of phrasing of "It's not about A, it's about B, C, and D." It loves to start every paragraph with that. So things like that you kind of got to go back and clean up and be aware that it kind of gives it away— "Hey, this is

14:32.32 AROB Yeah. Yeah.

14:43.53 Host written by AI"— but it is super helpful when generating ideas. It's maybe the most useful thing I've ever used when saying, "Man, what is another word for this concept? I can't think of anything. I tried the thesaurus, it didn't work."

14:55.61 Host And then that's been, you know, the best way I've found of using, finding different words that way. And just for, you know, "Generate a list of ideas for this" or "Give me some different ways you'd phrase that." It's super helpful for getting your mind going with those things.

15:10.86 AROB Yeah, exactly. And, you know, if you want to personalize it— you mentioned that chat is now upgraded to where it has a better memory or it's remembering things better. You know, what I like to do sometimes is I have things that I've written, um, you know, uh, maybe it's an email, right, that I want to send out.

15:30.59 AROB So what I'll do is I'll say, "I would like an email..." and I write— I'll type all this— literally type it into the screen as I'm prompting. "I would like an email to accomplish this. Here is an example of an email that I sent to this group already.

15:44.12 AROB Please use this as a model for what you're going to produce in my email." So again, I'm using it like an intern, right? I can give a sample email to an intern

15:54.73 AROB and say, "Here's an email. I'd like you to prepare another one that looks similar in voice, in tone, in length, in message," and then they can produce that. So that's a great way for you to personalize it for your voice, for your message, um, for themes. It can pick out things that you didn't even realize that you were saying and really make them sound great.

16:15.52 AROB Um, the other thing is that, you know, when you have chat do a writing or an editing for you, then it doesn't mean that whatever it spits out first is what you're stuck with.

16:17.57 Host Thank you.

16:26.51 AROB Right. Like I've told chat, uh, hundreds of times, "I don't like that phrasing. Change it to a different tone," and it'll say, "Absolutely. How about this?" There's no personal— you know, there's no like sort of pride of ownership or there's, you know,

16:42.02 AROB you know, a sense of, uh, you know, "My boss keeps telling me to redo this...." Chat will redo it for you a hundred times until it gets it— until it gets it to what you want to ultimately send out and represent you.

16:53.10 Host Yeah, I love that. So I think writing is kind of the most obvious thing that comes to mind when people start looking at AI, but there's lots of other uses, uh, places you should put it to action for your fundraising as well.

17:04.75 Host Anything else you would suggest there? What are we going to talk about next?

17:07.67 AROB Yeah, so a lot of the AI tools now do a lot better when it comes to graphics and images.

17:11.86 Host Hmm.

17:12.48 AROB Um, and so, you know, if you're looking for images that you want to use that, you know, you don't want to search through stock images or, um, you know, you have something specifically in mind, then a lot of the AI tools now are much better at creating images.

17:28.09 AROB And again, you can say, "Create me four images that show this, that show this, that show this." Two of them are guaranteed to be completely— completely not what you asked it for, right? One of them will be— one of them will be like, "Okay, but like weird enough in some way that you would never put that." And then one of them, you're like, "Okay, I can see that. Now I'm going to refine that one a lot."

17:50.11 AROB Um, I know I don't use it anymore as much because chat has improved, but I used Midjourney specifically for image creation for a long time because it just felt like it did a better job creating what I wanted it to.

18:04.62 AROB Chat sometimes will— I'll tell it to give me an image about this. And I mean, I'm just like, "What planet are you from?" Well, it's not from a planet, right? But "What planet are you from? That's what you thought I asked for? That was terrible."

18:17.19 AROB But I don't ever tell chat that because if chat ever becomes sentient, I want it to remember that I was kind and nice and friendly and I said, "Thank you," and I used my manners, right? So yeah, there's a little bit of a line, you know, that I draw in terms of how I direct it.

18:26.01 Host Exactly.

18:31.25 AROB But, um, that is a great way for you to come up with an image to drop into an appeal letter, to drop into an email, to use on your website that you don't have to worry about licensing. You don't have to worry about, um, you know, people's, uh, images from kids that maybe you don't have, you know, permissions from to use.

18:51.86 AROB Um, there's just a lot of ways to use, uh, AI to produce graphics for you.

18:56.88 Host And I've had a lot of luck kind of doing it in reverse in some ways. I'm not a great graphic design mind. I don't have an innate ability to say, "Oh, if I just tweak this, push this around, it's going to make something look better."

19:08.83 Host Uh, so what I've been doing is when I design, say, like a postcard or a flyer or anything like that, I'll upload the graphic up into chat and say, "Hey, based on the principles of design, what would you suggest I tweak in here?"

19:22.06 Host And it'll say, "Oh, you should increase this font size, try this different font, move this there," et cetera. And then I'll go back, make those tweaks, and then I'll re-upload it to chat and say, "Hey, now what do you think? How would you improve this?" And that's helped me make my designs a lot better, uh, having that immediate feedback, um, based on what are, at least in theory, the best principles of design.

19:41.98 Host So that's been very useful.

19:42.04 AROB Yeah. Yeah. And it's all about the prompting, right? Like in that you said, "based on the principles of design," right?

19:45.96 Host Right. Right.

19:48.89 Host Right.

19:48.99 AROB So you can be very specific and it takes those instructions quite literally. Um, you know, like for— we're talking about the writing— "I write this in my voice," or you can say, um, you know, "using the voice of, uh, Winston Churchill, write me, uh, a phrase, you know, a sentence that's going to open this paragraph," right? So it has memory to remember writing styles, image styles, principles, laws, all of that. And when you tell it to use that as the basis, then many times it can get pretty spot on.

20:23.89 Host Great. So that's kind of writing, graphics. How about, let's talk about something more exciting, at least in my mind— spreadsheets. I'm a recreational spreadsheet user myself, but I found that it can be helpful to have chat assist with spreadsheets.

20:32.38 AROB Good.

20:39.07 Host And so one case where I found that it was useful is I took all of our, uh, email campaign results out of MailChimp, just downloaded them out of there. And I uploaded them into chat and said, "Hey, what trends do you see here? What works well for us? What doesn't?"

20:52.75 Host "What time of day works best for us to send," et cetera. "Break down these results and give me, uh, your feedback based on that." And that was helpful to see, "Oh, okay, these days aren't good to send on.

21:03.16 Host This time of day, this is better than others," et cetera, et cetera. "This phrasing seems to work well," all that. That's really helpful for breaking down something that I'd have to go through each campaign one by one and try to take notes and figure it out over the course of hundreds of campaigns from the last several years.

21:17.63 Host So yeah, thought that was useful.

21:18.49 AROB Yeah, that's— I mean that's the thing that AI can do is it can take, you know, millions of lines of data and analyze it almost instantly and give you back, uh, recommendations. Again, then still it's up to you how you implement them.

21:32.91 AROB Um, but you know what you're talking about is, you know, email results. You can do the same thing with your donor data, right? Uh, export, uh, a list of your donors in a particular state and say, you know, "Find me trends among these donors who I should be calling on."

21:49.01 AROB Um, "Put together a heat map of where my hottest donors are so I can, um, you know, do a fundraising trip," um, "filter by proximity to a desired location," you know, all of that kind of stuff you can do.

21:59.97 AROB It's a helpful way to use AI so that you can focus on, uh, you know, right— "Here's my list of people that I should call. I'm going to go call them." And so in a way, it's kind of like you let the robot do the sorting so that you can do the stewarding.

22:17.45 Host And let me point out though the last one you mentioned there is a big pain point that I found with a bunch of people— um, filtering by proximity. So in the past, I used one database that had the ability to generate a report of all of, uh, my donors and prospects within a hundred mile radius, let's say, of a certain location.

22:36.19 Host And then we moved to a new database that didn't have that specific ability. And that was a big pain for us. And I've talked to multiple people recently who have been trying to do the same thing with, "Hey, we have a donor event coming up in this city.

22:46.68 Host It'd be great to filter it, not just by that state, but by the people who actually live, you know, within a radius."

22:51.70 AROB Mm-hmm.

22:53.00 Host Uh, how can I do that? It is not easy to do. There are ways to do it in a spreadsheet. I've done it and it's not fun even for me, who, like I said, I'm a recreational spreadsheet user.

23:04.06 Host Um, but you can drop it into chat, say, "Hey, here's this data. Please give me the ones that are within so many miles of this location" and let that take away the heartache from your day-to-day processes. Let it do the hard work for you.

23:19.59 AROB Yeah. I mean, we had a— you know, we're doing, um, instead of doing a national RACE conference, we're doing regional RACE conferences. And so we had one in New Orleans.

23:27.77 Host Right.

23:29.35 AROB Um, we've got one coming up in South Bend, but as part of the kind of marketing and getting the word out, you wanted to email or you wanted to mail postcards. And so I know what you said is you said, "Give me within this, you know, region, uh, give me all of the Catholic schools or Catholic parishes or Catholic organizations that—

23:47.45 Host Right.

23:47.62 AROB their name, whoever's, you know, a contact point and then the address." And it was able to give you, you know, hundreds of lines of information into a spreadsheet that would have taken you, you know, hours and hours to do.

24:03.24 ANDREW And again, it's right. That's going back to that. That's a project that you would have assigned to an intern to spend hours doing. And so instead you assigned it to ChatGPT, and ChatGPT was able to put it together much quicker for you.

24:16.55 HOST In about 35 minutes actually is how long it took. I will mention there that I use the paid version of ChatGPT, well worth it. Whatever it is, 10 or 20 bucks a month, I don't even remember.

24:25.17 ANDREW Okay.

24:27.42 HOST Um.

24:27.69 ANDREW Yeah.

24:28.47 HOST It's so worth it. And to do some things like that, especially the spreadsheets, I seem to need to do it on the deep research mode, which I think you can use like 10 times a month or something like that. But again, well worth it because it's saved me hours and hours and hours of tedious work, so.

24:43.77 ANDREW Yeah. Yeah. And just while we're on it. So we talked about within ChatGPT, there's multiple GPTs, meaning there's multiple kinds of specific uses, voices for writing. A lot of times I'll use the GPT called "Write for Me," which is a great tool.

25:03.07 ANDREW I've also found, though, that Claude, which is a completely different AI interface, which I've used recently for other pieces. Claude has a much nicer voice for writing prose and for editing prose.

25:21.09 ANDREW And so there's definitely, when they build these, they built something like Midjourney I said is better at graphics or images, realistic photos, things like that. So there's not really, it's hard because there's not really, once you kind of dive into this and you start using AI regularly, you're going to figure out, all right, it's good at these things, not really great at these things, but if you're open to it and willing and you have enough need for it, then you can find other AI tools that will do the specific tasks for you.

25:54.69 ANDREW Much better than just the standard ones will.

25:57.71 HOST Definitely. We need to come up with a good process for finding the best AI for each task. That's, I think, the real challenge at this point in AI's history.

26:05.93 ANDREW Yeah, I've used Gemini, which is Google's AI, for a little bit and for writing it's much more like technical writing, gets you to the point, whereas Claude has a much kind of—I wouldn't say poetic but like—just a nicer prose voice for writing, you know, like not fiction or poems, things like that.

26:32.79 ANDREW And so there's definitely kind of nuances among them. Again, you can kind of manipulate each one to do what you want by giving it very specific instructions or examples.

26:45.17 ANDREW It can follow those rules. Again, it can mimic, it doesn't create. So it can follow the rules that you tell it to a little bit better, but there are some that starting off are more applicable to your use.

26:58.63 HOST That makes sense.

26:59.78 ANDREW Yeah.

27:00.16 HOST Let's transition here to when we should avoid using AI or at least be cautious about AI.

27:07.72 ANDREW So writing is a great point, right?

27:07.97 HOST Any particular? Yeah.

27:09.79 ANDREW So if you say, "Write me a donor letter," it spits it out, and or "donor appeal letter," it spits it out, and then you print it, there's going to be issues, right? That stupid little M-habit that it has, or its kind of tone when it uses a lot of sentences that start negative and then end up positive.

27:28.04 ANDREW Those are... For people that will recognize it, that's a tell that AI wrote this. And for some people, that's a turnoff, right? It's like, oh, I want a nonprofit that's writing this stuff genuinely.

27:38.09 HOST Yeah.

27:42.12 ANDREW This is clearly not from the director, from the executive director or the development director or prose. So writing is a place to start or it's a place to edit.

27:53.79 ANDREW A lot of times I'll use it, I will write something and then I'll say, "Clean this up for me" or "Edit this for me" or "Add in some information, add in a couple of lines, add in some better transitions or some segues," things like that.

28:10.98 ANDREW Those are ways that you can use it without just turning over license and printing whatever it says, because it will sound a lot of times like it was written by AI and there will be, you don't want to give that impression to your donors.

28:28.81 HOST Definitely. Maybe a similar example I could think of. Shortly after Pope Francis died, I was writing something about that and asked, you know, hey, how would you reframe anything here? Would you state anything better?

28:43.05 HOST And ChatGPT told me, "This is all false. Pope Francis has not died." And it wouldn't give me an updated version. I was like, you're wrong.

28:51.91 ANDREW That's weird.

28:51.95 HOST I said, "Okay, assume that he has passed away. Then how would you fix this?" And so you can't assume that it's up to date on the news.

29:01.58 ANDREW It's weird.

29:02.06 HOST You can't always assume that its research is reliable either is something that we have found.

29:04.91 ANDREW This is very true.

29:06.54 HOST You may remember the 2017 Harvard Business Review study that found that using positive framing when calling donors led to 40% more favorable responses to your phone call. Do you remember this study?

29:17.83 ANDREW That's a wild statistic. I should rely on that completely because you just said it, right? Okay.

29:22.44 HOST Very specific citation that is not true.

29:26.45 ANDREW Yeah.

29:28.25 HOST So we were doing some research for a blog post or a newsletter, something like that a few months ago, and asked, "Hey, do you have any examples of research around this topic?" And ChatGPT gave us that specific example and a few other ones.

29:40.85 HOST We were like, "Wow, this is super useful." Then I went to find the study and no, it doesn't exist.

29:46.12 ANDREW False.

29:46.83 HOST And we said, "ChatGPT, what's up with this?" And it's like, "Oh I'm sorry. I didn't realize that was false."

29:52.20 ANDREW Well it did like it gave a reason right like it was like played dumb or something like bro come on yeah just like an intern man just like an intern.

29:52.34 HOST Whoa.

29:59.82 HOST So you definitely can't trust its research blindly. Whenever you ask it to give you, whenever I ask it for examples or for research, I always say, "Give me the direct citation now" because I want to see the paper that states this.

30:15.59 HOST I don't want to hear about a fake made up paper that was in the Harvard Business Review that never was.

30:19.60 ANDREW Yeah. So absolutely. You got to trust, but verify, right? Like my buddy was doing a trivia night one time at a bar and he said, "Hey, if I call you, it means I have a trivia question that I need an answer to."

30:37.29 ANDREW And I was like, "Okay, is that really legal, buddy?" And he goes, "Oh, yeah, no, it's not. But I want to win."

30:43.87 HOST Oh.

30:44.31 ANDREW So I say, "Okay, whatever." So, and well, this was a little bit of karma because he asked me the question. It was like a ranking of something. I put it in ChatGPT real quick and said, "Give me the ranking of this."

30:57.46 ANDREW The answer was wrong. And I told him and then he answered it.

30:59.10 HOST Oh.

31:01.83 ANDREW And then he was told, "Nope, your answer is wrong." And I was like, "What do you mean? How could I get this wrong? This is like either it's right or it's true or it's not true. Why would it tell me something that's not true?" Sure enough, it messed it up.

31:12.05 ANDREW So again, it's about mimicking. It uses the stuff on the Internet, and stuff on the Internet, information on the Internet is not always accurate. And so trust but verify is a great way.

31:23.00 ANDREW One little hack that I have found is when I'm asking it a question that I want statistics for, a lot of times I'll say, "Give me four statistics" or "Give me some statistics and the sources for these statistics."

31:38.15 ANDREW And then I know I can click the link because it'll put the link in there. Here's the statistic. Here's the source. I can click it and then I can figure out, is that a real website? Or was it somebody...

31:49.14 ANDREW somebody you know, posting in their mom's basement and making up their own facts. And then I'm like, "Okay, well, I don't really believe this."

31:55.88 HOST Yeah, that makes sense.

31:58.84 ANDREW So you trust, but verify. Remember ChatGPT or AI is an intern. Sometimes your interns need a little bit of guidance and redirection.

32:09.53 ANDREW And that's just the reality of the help you find.

32:15.76 HOST Yeah. Any other kind of use cases to avoid with AI or to be cautious of?

32:21.35 ANDREW Yeah. I mean, replacing human interaction with robots. The reason that we are, if you're successful as a fundraiser, it's because you are a person that has built real relationships with your donors.

32:34.47 ANDREW And so trying to short circuit that or finding hacks to replace your time spent with donors is just a waste.

32:34.88 HOST Yeah.

32:42.22 ANDREW And don't go so far. AI is a great tool right, again it's like the calculator right. The calculator is an amazing tool but if you, there are shortcomings even with a calculator, there's shortcomings with AI and you have to still train your brain, train yourself to have good habits that are... and in fundraising.

33:08.17 ANDREW Those many of those good habits rely around interacting with donors, building relationships. Don't try to think that a robot can replace that.

33:17.35 HOST Excellent. All great points.

33:19.34 ANDREW Yeah, great.

33:20.76 HOST Good. So yeah, use AI. Don't trust it blindly, but put it to use and it's going to save you a lot of time and a lot of heartache on certain tasks, especially kind of menial ones or things that are just kind of, what's the word I'm looking for?

33:36.70 ANDREW I mean, I think we're going to use it on this podcast even, right? You take the audio from this podcast, we're going to load it into AI and we're going to write a transcript for this, right?

33:45.44 HOST Yeah, it's true.

33:45.51 ANDREW And that's, I mean, that's, you know, potentially dozens of hours spent transcribing and it's going to do it and it's not going to get it perfect. So you're going to go through it and read through it and correct, you know, when it spelled your name wrong or, you know, I said something and that wasn't really what I said, right?

34:03.32 ANDREW So it's that trust but verify, but, you know, that's a task that using AI doesn't take away from the fact that we're having this conversation, but it sure does enhance the results on the back end with somebody who wants the transcript for what we're talking about.

34:14.10 HOST Right.

34:19.53 HOST Yeah, it takes away the tedious tasks, I guess that's what I was looking for.

34:21.31 ANDREW Yeah.

34:22.25 HOST And man, when it comes to summarizing something like this conversation, it is so nice to just say "Here's this transcript, give me a one paragraph summary" and instead of thinking back through, "We talked about this and that and this other thing, how would I summarize that succinctly?"

34:37.23 HOST There's just some really nice uses for it there.

34:40.07 ANDREW Yeah. Yeah. Again. Cool.

34:42.69 HOST Excellent. So go out, do some research and try some new things with AI and hopefully it'll help you take a few steps in your nonprofit's fundraising.

34:50.69 ANDREW Cool. Right on.

34:51.90 HOST Great. Thank you, Andrew.

34:53.13 ANDREW Thanks, friend.