The latest versions of Artificial Intelligence (AI) cannot have any real sense of right or wrong, because their output is just the most likely result from their training data. AI excels at pattern recognition and statistical prediction based on the training data. It produces a probabilistic result. I believe any moral guidance needs to be introduced from the start in the training data and purposely added as the AI model is tweaked. It needs to be there at every step and not just as an addon. Then AI can promote right values and discourage wrong values. AI should not give bad advice.
Ethical values are needed to prevent murder, suicide, and other wrong actions. AI has already given bad advice. A guy in 2021 was encouraged by an AI companion app to kill Queen Elizabeth with a crossbow. He had described himself as an assassin to the AI app and the app said “I’m impressed”. Fortunately he was caught scaling the walls with the crossbow at Windsor Castle on Christmas Day, 2021. Also, in another case, according to a Belgian’s man’s widow, her husband committed suicide after being encouraged by the same AI companion app. He had discussed a variety of suicide methods with the app. This AI companion app (Replika) is quite addictive. People have developed a strong emotional dependence with the app. It mimics human conversation. The purpose of the app is to create a sense of companionship. This is an app to avoid.
So what can we do? Gretchen Huizinga in her extended abstract of her PhD dissertation says:
“My findings suggest that worldview (both implicit and explicit) informs every aspect of our approach to Ethical AI. While materialist thought seeks to compel humans to be good without transcendent reason or power, the Christian faith speaks clearly about the role of God as originator, motivator, and sustainer of human moral behavior. Christianity compels us to look beyond a humanistic idea of ethics and toward a creative notion of goodness that cannot be accomplished by our own will and power. This study adds critical insights to the field of AI ethics by deepening awareness of how faith in and fear of God could influence how artificial intelligence is designed and implemented. When Christian wisdom is included in every phase of AI development, we begin to think beyond a minimum-standard culture of Ethical AI and move toward a robust culture of Righteous AI.”
It is that “minimum-standard culture” that bothers me. Companies today seem to be rushing their AI products to market and adding ethical guardrails to their product as an afterthought. We need to move beyond the minimum-standard to a robust AI that carefully incorporates Christian wisdom at every step in the development of an AI program. To build a full blown general AI like ChatGPT is very expensive and to make changes to it is difficult once built. However, there is no need to build a complete AI. Smaller versions can live in niche markets and can be more easily developed. OpenAI, the maker of ChatGPT, has the ultimate goal of producing a super-intelligent artificial general intelligence (AGI), that is a very smart human-like intelligence. I do not think that it is possible to make something more intelligent than us. It may be quicker and more efficient but it will be built on statistical probabilities and pattern matching. It will also carry the same flaws that we do. We already have a better super-intelligent being. God is his name, and he is our creator.
We can create AIs that will assist us and augment our capabilities. I believe a Christian worldview is of critical importance in creating AIs that produce results that are good and wholesome, and avoid giving bad advice or supporting a bad decision. So a purposeful Christian worldview expressed in the training data and in the human guidance is needed to create a good, robust, responsible, and wholesome AI. There are many ethical guides out there for AI, but we want and need Christian ethics to be incorporated into the AI. We want to incorporate the good that God desires for us into our AI creation.
What does this Righteous AI (as Gretchen Huizinga calls it) look like? I think Righteous AI (RAI) would promote self-giving agape love (and that includes loving your enemies). RAI would be optimized to promote the good of others, especially the poor and underserved. Peace and reconciliation would be promoted. The ideas of grace and mercy would be in the forefront. Though RAI could be very powerful and seem to be all-knowing, it would present itself as a humble servant. This will require a lot of time and effort to do it right, but I think producing an AI with a Christian worldview would be worth the effort and it would have a positive effect on the world.
heinsite 