The AI industry’s north star is Artificial Superintelligence (ASI). The goal is to invent an entity that vastly surpasses human capabilities, or in more blunt terms, a god. Contrary to popular narrative, it has never been the case that we could drop a bunch of data into a neural net, and the algorithm takes it from there. Current AI systems, presently dominated by large language models (LLMs), are shaped by human decisions. If ASI ever comes to fruition, it will be covered in human fingerprints.
If we inspect how generative AI systems are engineered, humans have influenced them at every turn. We have authored all the training data, mathematics, and code. We have determined how to clean and represent the data, and which data we ought to exclude. We have fixed or optimized the hyperparameters of the neural nets. We have embedded human preferences into how the models respond to us. We have determined how to evaluate the models. We have wrapped these models in nice interfaces that we have designed. The list continues.
AI is not magic nor an act of nature – it is a human creation. The marketing that pervades AI promotes the former. If we instead classify AI as a system that is carefully crafted by humans, we might ask ourselves an interesting question: can we engineer our way to ASI? If AI is believed to be “a magical intelligence in the sky”, we might think ASI is imminent, perhaps only a little more data is needed. However, when we think about AI as a human-engineered system, this should give us caution about how looming ASI is. Humans will have to be creative and inventive if we are to ever get to ASI. It will not be like an archaeological find, it will be born out of what we are able to devise. Even the mechanisms and paths for recursive self-improvement are designed by humans. An LLM will not magically figure out these strategies without direction. Likewise, the “emergence” we observe in neural nets is directly born from associations in data that we created and from leveraging mathematical algorithms we designed.
Technological limits aside, would we want a god-like system expressly derived from the complex good-evil paradigm of humanity? If ASI occurs, we clearly would not have discovered it; we would have invented it. The reality should put a larger weight of responsibility on those pursuing the goal of ASI. We cannot claim it arose of its own volition; we must take ownership for what was ideated, constructed, and nurtured.
As Christians, we believe God has designed the world. For me, Christian Creation is about why and not how. We know that His ways are not our ways (Isaiah 55:8), and His glory is evident everywhere we look (Romans 1:20). He is the Creator, not us (Psalm 8). At best, we can devise a derivative recreation of what God has authored. This could be a powerful creation, but one that is imprinted with our human flaws. For instance, stochastic gradient descent, an important equation for training some neural networks, does not create a comprehensive and clear world model. It will happily apply tried-and-true heuristics to arrive at answers, which sits in stark contrast with the intentionality we know God has endowed on His creation.
The single-mindedness of following the north star of ASI has another cost. The dogged pursuit of any objective, including ASI, can cause us to miss the wonder of how the world actually works. Oftentimes, the steps to achieve a goal do not resemble the end state. Babies must learn to fall before learning how to walk, though falling appears to be the opposite of walking. The same principles apply to the pursuit of ASI: the research community needs the humility to explore and wonder to achieve greatness.
God has created a broader design that we can explore and understand from our human vantage point. We should marvel at what God has created and value our role as beloved stewards. This viewpoint leads us to a place of reflection, searching, and groundedness rather than a place of believing we undoubtedly can know all the details to every question.
Knowing our place as humans has perhaps never been more important. I believe God has called us into partnership with Him. We are not equals, yet we are called to be in close relationship with our Maker. Our inherent desire for knowledge and control can turn into a single-minded obsession. God simply wants us to seek Him and embrace who He created us to be. The fixated pursuit of knowledge distracts us from experiences and relationships that God can use to sanctify us as part of His larger plan. We can certainly pursue ASI, but potentially at the cost of losing sight of what matters.
Views and opinions expressed by authors and editors are their own and do not necessarily reflect the view of AI and Faith or any of its leadership.
