As more nonprofits begin to implement AI for fundraising and other tasks it will be important to consider the ethical pitfalls of AI. Especially AI for fundraising where there is a lot of personal data in play.
If you haven’t had a chance to read our previous posts on AI, here’s a quick refresher. AI stands for artificial intelligence. AI is a computer or a machine that is capable of conducting tasks that require human intelligence.
Today’s blog is all about ethics and AI for fundraising. What are the ethics considerations that nonprofits need to keep in mind? And what can nonprofits do to avoid ethical pitfalls when implementing AI for fundraising.
Bias in the Source Data
One of the biggest opportunities that AI offers is the ability to analyze data and patterns in past behaviour to make predictions about future behaviour. When the data that is being analyzed is biased the AI will become biased as well.
Here’s an example, when Amazon tried to program an AI tool for recruitment, it inadvertently created a biased system. Looking at Amazon’s current roster of engineers the AI saw that most of them were male, so it excluded candidates with female sounding names. Based on the data the AI determined that males exclusively must be the best candidates.
Bias in Fundraising Data
Fundraising data can be biased for the following reasons:
- Historically the profile of a major gift donor has been androcentric and xenocentric
- Often fundraisers have sought out friends or acquaintances of board members or other leadership, these connections would reflect the diversity of this group, or the lack of it
- Wealth screening tools use sources like postal codes, which might not identify a wealthy individual who lives in a more modest neighbourhood
The Effects of Bias on Fundraising AI
If you use this source data to ask fundraising AI “who are my best donors based on previous major donors”, you will get a narrow profile and this isn’t the outcome that you want the AI to deliver!
As we are working to build a more inclusive and diverse nonprofit sector, as fundraisers we want to make sure that our major gift pipelines represent those values. Working with a narrow prospect profile also leaves money on the table. You will be missing valuable prospects who don’t meet the profile.
The “Curse” of Knowledge
Ever been creeped out by a re-marketing ad on the internet after you browsed through an online store? Imagine how you’d feel if AI was able to predict highly personal things about you. That’s exactly what happened to a teenager in the United States.
Target uses an AI predictive tool to analyze the shopping habits of customers and predict what they will need next, or predict major life changes like pregnancy. Based on a teenage girl’s shopping habits this AI figured out that she was pregnant and she was sent coupons for baby items.
When her father saw this he was upset and complained to the store. But the AI was right, his daughter was pregnant.
Data Collection for Nonprofits and AI
Here are a few cases where AI could be used to collect too much data:
- Facial recognition can be used to see how attended an event
- And rather than look at who attended their own events, a nonprofit could use this technology on a competitor to find out what people attended their event to pursue those donors
- AI could be used to scrape annual reports and digital donor walls of other charities to put together a financial profile
At what point does the data collection become unethical, especially when it comes to using the data of other nonprofits?
The Effect of “Knowing Too Much” for Nonprofits
The nonprofit sector will need to determine what data collection is ethical or not. The guide post to answering this question needs to be whether or not donors would be comfortable with the amount of information we can collect? If there was a data breach and this information was leaked, does that pose a threat to donors?
Donors who feel creeped out or those whose security is compromised by your data practices won’t want to support your nonprofit and might stop donating to any nonprofit in the future. So these questions must be carefully considered.
What Can We Do?
There are many opportunities for ethical pitfalls with AI for fundraising, but that doesn’t mean that AI is bad and we need to avoid it. AI is already very present in our lives and it can be a powerful tool, just like anything it can have unintended consequences.
As fundraising AI continues to develop, the nonprofit sector needs to have conversations about ethics.
- What are the ethical pitfalls of AI for fundraising?
- What would our donors be comfortable with, what might make them uncomfortable about using AI for fundraising?
- Conversations with AI vendors during implementation about ethics and how vendors approach ethical questions.
This open dialogue will help to ensure that AI continues to be a powerful tool for nonprofits without unintended ethical consequences.