Can We Remain Friends While I Use AI To Predict Your Behavior?
I am not against AI - I use it. I think it holds great promise in many areas. However, I am concerned about its ethics and impact on development, fundraisers, donors, and small organizations which could be left behind when AI is embraced by mid-size to larger institutions.
Before I share my ethical concerns about AI within Development, I will provide background information on predictive and generative AI. Understanding the basics of AI is important when weighing the pros and cons of the tool. I will also outline potential ethical challenges. In conclusion
I will provide a path forward for the ethical use of AI that respects both the science and art of Development.
What is AI?
There are two basic forms of AI ~ Predictive and Generative.
Predictive AI is able to collect, sort and make estimations about the future.
Generative AI is able to create new written content, music, and design.
There are common activities / elements but differences between Predictive and Generative AI.
In most cases, when AI is or will be used with development efforts, both Predictive and Generative capabilities are used – and each has its own set of risks.
Ethical Issues with Predictive AI
Advances with Predictive AI are both rapid and significant – the faster it moves the bigger each stride becomes. Below are six potential ethical issues that may need to be considered when using Predictive AI:
Biased Info-In-Biased Info Out: One major concern with predictive AI is the potential for bias and discrimination in decision-making. If data is biased, or if algorithms are focused incorrectly, a discriminatory outcome may result.
Lack of Transparency: The users of Predictive AI need to understand how the inner workings of their AI solution function. Understanding the inner workings will help users avoid discriminatory outcomes through potential biases and errors.
Privacy Concerns: Predictive AI systems often require large amounts of personal data in order to make accurate predictions. This can raise privacy concerns, especially if the data is not properly vetted and protected.
Human Oversight: Predictive AI can influence decisions so proper oversight must be used to eliminate unintended consequences and potentially harmful outcomes.
Automation Bias: People tend to trust machines – think of an MRI machine. We accept their accuracy without question. Therefore, people must know for certain the algorithms are accurate and free of biases.
Accountability: A good starting point when considering AI is how people will be held accountable for its use. If this is not clear and there is a bias towards automation, problems could result.
Generative AI
Similar to Predictive AI, Generative AI, is growing at an exponential rate. Below are four risks of Generative AI:
Data Privacy: Generative AI draws from mammoth data sets. It is not beyond reason to think that some of that information might be private and the data could be accidently shared without a person’s consent.
Rights to Intellectual Property: Generative AI is able to produce images that are extremely close to other creations.
Will AI be the Tide that Raises All Boats: Generative AI has the potential to transform development. Will that benefit finds its way to the small to mid-sized organizations? Will they be able to afford it? It will be interesting to see if all development offices will benefit or just the larger and wealthier institutions.
Human Touch: If AI decreases the amount of time needed to know a donor it will not likely be long before the average donor portfolio size increases – which may lead to less one-on-one time with donors since less research will be done manually.
In considering these ethical issues, it's vital to remember the people behind the statistics—the impacted individuals and communities. The development and deployment of generative AI must be guided by an approach that considers the various risks.
While Predictive and Generative AI may have different purposes and methods, they both have the potential to greatly impact the development profession. Ignoring or choosing not to utilize the technology is not prudent as it can be of great assistance. However, a path needs to be identified to navigate the potential ethical hazards.
Is there a middle ground?
I work in the area of human services. I do so because I want to see clients experience their best life.
I also want to see that in others – and this includes donors. Charitable giving is important not just for the organization but also for the donor. Asking for support and giving brings us “out of ourselves” and into living for others.
The basis for good relationships between a development professional and a donor is not AI. It is a type of relationship – friendship. There are different types of friendship which are segmented by the type of donor with which you are working.
A 2,300-Year-Old Solution for AI in Development?
The basis of all development is relationship and friendship but not all relationships and friendships are the same. Aristotle taught three kinds of friendship ~ utility, pleasure, and virtue. These three categories can be used as a guide to how AI can be used whilst raising funds.
Utility is a relationship based on what others can do for you or what you can do for them. This may include acts of kindness that are returned by a gift or favor. These friendships end with the interaction and do not directly influence one’s character.
Pleasure is based in shared activities, passing pleasures, and good feelings. Think of a shared hobby or time-bound cause commitment. These friendships wax and wane as quickly as the common interests that brought the individuals together.
Virtue is a type of relationship wherein you befriend people for who they are, how they influence you, and how they encourage you to grow. These relationships require people to be at the same “level,” and are stable friendships and not fleeting.
While Aristotle held the friendship of virtue as the highest level of friendship it is also the most challenging. To know someone requires truthfulness and the willingness to trust and be trusted. It is here, in this sort of relationship, where two equals may come together for a cause and share the benefit over time. In the end, each person is in the relationship for the good of the other.
The foundation of Development and especially Development for Major Gifts must be rooted in a virtuous relationship where the fundraiser and the donor are each looking out for one another as the basis of the relationship. This means they are equal…but in different ways that are complimentary; for example, part of a donor’s desire might be for recognition when making a substantial gift and the development officer has the ability to facilitate that recognition. From the negative this could be viewed as transactional. From the positive it can be viewed as two people in a relationship helping one another. They are equals in the sense that both bring an offering “to the table” that benefits the other.
Part of these relationships is getting to know one another. It is through dialog that this information has traditionally surfaced. However, when asking a donor for money can the development professional still present themselves as “equals” if they are using Predictive AI? There is a difference between Generative AI (wealth screening) and Predictive AI that informs development officers to take specific actions. In short, can one enter a relationship encounter when they “know” what a person, through the sophisticated use of algorithms, might do? Can this relationship still be considered a relationship of virtue? A rationale needs to exist to wade through these unknown waters.
AI Governor
Have you ever seen a moving rental truck, with a sticker that says the truck cannot drive more than 55mph? These trucks are equipped with “governors” that limit the speed of the truck to help keep it under control. The need for a governor recognizes the realities of the truck needing to move but not at an imprudent speed. Likewise, some organizations could find its Development efforts fall behind when other organizations adopt AI for development. It is likely that not every development office will be able to afford AI.
Governors In Action
The path forward, in a way, reviews the three types of relationships outlined by Aristotle. I am arguing that there is a need for “governors” at various levels based upon the type of relationship between the donor and the development professional. More to the point, there are times when a development office is working with “cohorts” and other times when they are working with “individuals”. By their nature these different types of relationships require different actions of intent, in addition to basic security measures that are afforded to all.
A chart of examples may be helpful:
Challenges with this Model
The above model has a few challenges to consider.
To begin, the relationship definitions are driven, to a certain extent, by the level of giving which could give the appearance that the institution only cares for money and not the donors. On the other hand, money is, in fact, that primary end of development efforts in addition to building trust and affection for a specific institution. It’s measurable.
At the “lower level relationships” the governor is set higher because the development officer is working more with cohorts than individual people. The level of customization tends to be higher and neither party is as vested in the relationship. As a result, there is less responsibility by the institution to nearly or non-existent relationship where high turnover is the norm. What the above model does not do is “set the levels” between the types of relationships. In the end, these donors, tend to be in high number and working with them as a cohort makes more sense.
The idea of dividing relationships as suggested, and especially the nature of the highest levels of relationships, may seem pollyannish. However, there needs to be a starting point and the starting point should be on the foundation of development which is relationship.
Conclusion
Both Predictive and Generative AI present the field of development with a host of benefits for institutions and benefactors. The ability, for example, to highly customize letters in direct mail campaigns will be of great assistance to all. The ability to build more effective and impactful donor portfolios will help giving, build relationships, and even help on practical matters such as reducing travel.
The ethical challenges that AI produces can be reduced through thoughtful policy, more focus on relationships, and constant renewal efforts.