Chat for Blackbaud AI: Spread the Word!

I would tell them to give it a shot, but that it's not ready for prime time. It frequently makes mistakes and does the opposite of what is asked. Just this morning I asked for a list of 10 donors who have been highly engaged in the past but have not given this fiscal year. It came back with a list of donors who have all given this fiscal year. I asked it to correct it, it said sorry, and then proceeded to give me the same incorrect list again.

Right now, I would tell them not to use it. In fact, they can't use it because I haven't turned it on for anyone but myself. I test it every so often, and the answers it gives are always wrong in some way. I'll happily turn it on for everyone once it's working well.

I’m rooting for improvement, but right now I’m not recommending it to most of my clients. During the DBV‑to‑WV transition, they already have so much on their plates, and the tool’s accuracy just isn’t consistent enough yet. In many cases, the time it takes to sort out what Chat for Blackbaud AI got wrong versus what it got right is longer than simply running a report or looking up the information directly.

At this stage, I find it’s most helpful for polishing donor‑facing language—not for analyzing or interpreting data from the database. I’m hopeful that will change as the tool matures.

I enabled AI for one of our more techie development officers. So far, he said it will say it can do things for him like creating a specific list but then is not able to. Very mixed results. He said he's willing to keep 'playing' with it.