Zurück
,

Not a Black Box: The Recipe Book for our AI

StiftungSchweiz has been using AI assistants for a few weeks now and is taking a bold step: we are publishing all our rulebooks transparently. You can find out why we do this and the ethical considerations behind it in this article.

The use of artificial intelligence (AI) in the philanthropic sector presents many opportunities, but also some challenges. StiftungSchweiz has had some clever AI assistants on board for a few weeks now. These should be used responsibly and sensibly, with security and transparency having top priority. All developments take place on Peakprivacy.ch, Switzerland’s secure AI platform.

Focus on Ethics

How do you make an algorithm simple and easy to understand? The key to this is the rulebook. When we develop AI tools, we write down the rules that control the AI in such a rulebook. From a technical point of view, an AI that is permanently implemented on a platform is not a simple chat prompt such as GPT, but a more complex system such as RAGs (Retrieval Augmented Generation models) or Agentic Networks. Our technicians therefore divide the rulebook into digestible chunks to create a stable system that always answers the same question in a similar way.

In a way, the rulebook is the recipe for an AI assistant and represents valuable intellectual property. In addition to technical sophistication, it also contains in-depth philanthropic knowledge that we use in consulting, for example. Nevertheless, we have decided to publish all of our rulebooks transparently. This is because StiftungSchweiz is committed to transparent philanthropy, not a black box in which algorithms make decisions in a mysterious way.

Rulebooks: Feedback Welcome

But what can AI even do on stiftungschweiz.ch? Our cover letter generator creates a print-ready cover letter that cleverly bridges the gap between your project and the requested grant giving organization. The rulebook shows you that this is not rocket science, because cover letters are highly formalized types of text that must be structured precisely and correctly – brevity is the spice of life.

The project assessment applies best practice in the description of a project, it analyzes your project portrait and points out specific gaps or potential for optimization. What is a smart target? Does the desired impact relate to the problems of our present? Are the target groups defined precisely enough? By the way, we think that the wizard’s judgment is still a bit harsh – in the next version we will make it a bit more motivating, the new rulebook is already in the works.

Supportive, not Autonomous

Algorithmic decision-making raises key ethical questions. One of the biggest challenges is to ensure that the algorithms are fair and unbiased. The risks of unpublished algorithms are considerable: they can make decisions that are opaque and difficult to understand for those affected. By publishing our rulebooks, we want to counteract this and promote fair and equitable decision-making.

Have we piqued your interest? Take a look at the rulebooks published so far (more are being published all the time) and share your views with us. By being open and transparent about the algorithms we use, we are committed to the responsible and ethical use of AI in philanthropy. Because only through transparency and cooperation can we fully exploit the potential of AI and at the same time ensure that it is used for the good of society.

Bootcamps: AI in Philanthropy

StiftungSchweiz would like to speak openly about the algorithms we use on the platform. We do this in order to gain and maintain the trust of our users. At the same time, it is important to us to train as many informed stakeholders as possible who are aware of the potential and risks associated with this new technology.

As part of two boot camps for the target groups of Nonprofits (organizations seeking funding) and Funders (grant giving organizations) in mid-June and early September, we invite you to a condensed discussion of AI and philanthropy. How much hot air is there in the omnipresent discussion about AI and where are the rules of the game actually changing? Our compact bootcamp provides clear and tried-and-tested answers to such questions.

Das könnte Sie auch interessieren