Rome call for AI ethics

I serve as the Christian Reformed Church of North America’s representative on the Faith and Life Science Reference Group (FLSRG) of the Canadian Council of Churches. The FLSRG focuses on the relationship between science and our God-given humanity. Often, I get to learn about and discuss topics I normally wouldn’t delve into. We explore genetic engineering, the impact of Artificial Intelligence (AI) and, currently, how Medical Assistance in Dying (MAiD) is expanding in Canada.

More about AI

FLSRG has signed on to an initiative called the “Rome Call for AI Ethics,” a broad statement initiated largely by the Catholic Church and signed by various partners – churches, industries, universities, professional associations and governments. AI’s impact on us all is significant, and the Rome Call aims to provide a framework for its ethical development. The Call was drafted in 2020, and at our last FLSRG meeting, we discussed how the rapid development in AI over the last few years means an updated version seems necessary.

The Rome Call is based on three values. The first is ethics: “AI systems must be conceived, designed and implemented to serve and protect human beings and the environment in which they live.” The second value is education: all people must have access to and training in AI systems. Third is rights: we all have rights, which must be protected. These three values are then expanded into six ethical principles: transparency, inclusion, responsibility, impartiality, reliability, security and privacy. The Rome Call is commendable in its efforts to ensure AI does not harm us or our world.

Recent developments

Yet our committee sees areas where the Rome Call needs updating because AI is already much more complex than it was in 2020. There are AI programs designed to perform specific tasks better than humans. AlphaFold is an example; it predicts protein structures – something we humans can’t do on our own – and has revolutionized our understanding of the biology of proteins. Then there are generative AIs that can make new content based on prompts provided by the user, like ChatGPT. These large language models can often produce text and other items, such as images, that rival those produced by university students. They can summarize large bodies of text and distill the main themes. Unfortunately, generative AIs have also been shown to have biases, and they can create what have been called hallucinations – or, more simply, untruths.

A recent development is what are called AI agents, programs that make decisions autonomously in complex environments. One example is a program that can plan a variety of tasks, such as vacation itineraries, based on user prompts. A self-driving car is another example of an AI agent. While the ultimate potential of these AI agents is unclear, there are major ethical concerns, including cybercrime, data privacy and legal liability if things go wrong. An update to the Rome Call would need to address these issues.

Sinful robots?

We are made in the image of God, so it is not surprising that we have the creative ability to produce complex AI programs that can do many things. However, at the same time, we are fallen individuals, and our AI models share in our sinful tendencies. We need to protect ourselves against the sin that our creations cause. Pray for wisdom as the AI world develops.