AI and the "God of the gaps"

What the 'God of the gaps' fallacy teaches us about AI debates and our humanity.

The 'God of the gaps' concept is a fallacy. It asserts that scientific mysteries imply supernatural causes. If we cannot explain X scientifically, the argument goes, it means that God must be the cause of X.

Thoughtful theologians don’t usually argue for God in this way, understanding that as science advances, it explains more, making the gaps - and therefore God - smaller.

I won’t discuss how they argue for God instead, because this is not a religious newsletter. The reason I find the “God of the gaps” concept interesting is because it mirrors a similar fallacy in discussions about the advancement of AI.

Can your AI do [fill in the blank]?

As AI advances and becomes more powerful, there is growing concern about what makes us uniquely human. Just as some fear that science may leave no room for God, others worry that nothing about humanity will seem special if machines can mimic us completely.

This is a fallacy. Our humanity doesn’t come from being unique. It comes from being human.

The interesting question is this: if we can one day build a machine that is indistinguishable from a human, will it be human?

Keep pushing forward

The 'God of the gaps' fallacy wrongly claims that what science cannot explain is special, even supernatural. The same fallacy appears when we think that our humanity is special to the extent that today’s AI lack certain capabilities.

The key term missing here is 'yet': science has not yet explained everything; similarly, we have not yet built an AI with certain capabilities. Just as science will explain more, and fill more gaps, AI too will grow in its abilities.

Even otherwise smart people will sometimes confidently declare, “AI will never be able to do [fill in the blank]”. This is a mistake, and likely a reflection of fear and wishful thinking.

Like scientific gaps being filled, AI will eventually achieve what some may currently think impossible. Our humanity is in our behavior, not in what machines lack today. Whatever you think of machines more powerful and more creative than humans, it's wiser to prepare than hope it will never happen.