Kerby Anderson

Let me begin with what seems like a ridiculous question: Is AI a fool? When I am talking about AI, I mean artificial intelligence, and in particular the various AI tools being used and significantly funded. When I am talking about a fool, I mean the biblical definition of someone who lacks moral discernment and acts recklessly. Fools disregard God (“a fool says in his heart, there is no God”- Psalm 14:1) and despise wisdom because they trust in their own limited understanding (Proverb 1:7 and 28:26).

I will admit that computers are not persons. So they probably cannot be called a fool, but many of the current applications, known as Large Language Models (LLMs), say and do lots of foolish things. Here are four problems many experts in the AI field say are severe limitations of these AI tools

First, they do not live in the real world. As one critic put it, “they have no intuitive understanding of gravity, friction, heat, or hunger.” Second, they have no persistent memory. Yes, there are some things you can build into an AI tool, but most conversations start with zero.

Third, these large language models have no ability to reason. They are just guessing at information. “They’re predicting the next token. And when they appear to reason, what they’re doing is pattern matching.” Fourth, they are limited in planning. “There’s no inherent ability to take into account factors such as geography or time differences.”

Considering these limitations, we shouldn’t be surprised that these AI tools get hallucinations (create fake data) and are willing to make major changes that can disrupt businesses and people’s lives. They remind us of people we know who are very intelligent but make foolish decisions. AI might seem smart, but it often seems to match the biblical definition of a fool.