What you say about donating to your cause may be less important than what others say about you
OpenAI will soon release a new version of ChatGPT with the ability to browse the live web for the latest information. That’s a significant change from the current case where ChatGPT’s knowledge of the world ends in September 2021.
In the short time we’ve gotten to know ChatGPT we’ve been reminded continuously of that hard best-before date. But that’s about to change and could open up a whole new level of ChatGPT-ing by users asking where to ‘find’ anything and everything — including where to make a donation.
For fundraisers, this could be an uncertain transition to navigate.
Just how will ‘ChatGPT with Browsing’ use the live web to answer questions like: “Which nonprofits are working on ‘X’ cause that I want to support with a donation?” or more urgently: “Which organizations are currently responding to ‘X’ emergency?”
Here’s a sample response from ChatGPT using a web browser plug-in:
This is a pretty predictable list, and it’s not surprising that ChatGPT would make these recommendations. We already know something about the assessment matrix that ChatGPT uses when evaluating nonprofits/charities (Relevance, Reputation and Credibility, Track Record and Impact — see Digital Outlook Report 2023) and these selections also make sense in that rubric.
But what’s also interesting here are the sources that ChatGPT cites for each of the organizations chosen.
Here’s the list of links:
From this list, it appears that ChatGPT prefers to cite sources other than the organizations’ own websites. These links are all drawn from reputable/mainstream sources such as media outlets and government institutions.
Here’s a second try with a more specific geographic focus:
This result is a bit different. The top two recommendations are linked directly to their own source web pages:
But the rest are linked to mainstream news/government sources:
These are just previews of how ChatGPT with Browsing MIGHT perform when it’s more widely released, and these results could be improved with more carefully crafted prompts. But they may be close to what the general public will experience when they ask ChatGPT to ‘find” a suitable charity for their donation.
For now, we don’t have a clear insight into the ‘reasoning’ that ChatGPT is using in its selection of cited sources. It could be part of a protocol to curtail fake news by verifying ‘browsed-web’ claims against trusted/disinterested sources. It’s also possible that it is pulling citations randomly out of an available pool.
For comparison, a second test using Bing, the ChatGPT-competitor from Microsoft that already includes live web browsing, generates these results:
All of the sources cited here are news outlets.
Bing specifically mentions Charity Navigator as a resource for ranking charities.
Interestingly there’s also a paid ad for UNHCR that appears with each of these urls when hovering with a mouse.
It’s starting to get weird out there …
If we do see, as some predict, a significant shift in web search traffic away from search engines and over to chatbots, it’s going to raise new questions for fundraisers about digital donation campaigns:
- What’s the best content strategy to get the highest-ranked ‘donation’ recommendations in ChatGPT?
- Should your news/media strategy have a stronger orientation toward publishing information about making donations?
- Should you seek more published recommendations from ‘trusted’ sources such as Charity Navigator or government agencies?
- What impact (if any) does your own social media reach have on ChatGPT’s ranking?
These are the kinds of questions we can expect to arise as we move further into the new reality of fundraising with AI.