AI might be another Babylon tower

I want to start this year with something deeply philosphical. Something that might not be really comfy to talk about. Despite of that this sparked and therefore is eligible to appear in this blog.

Today I want to share my thoughts about Babylon tower and the AI hype we’re living in. If you’ve missed or forgot what Babylon tower is about, I’ll remind you briefly. That’s the part of the Bible, that describes how people attempted to build a Babel tower to reach the sky. On one day people who were doing pretty well - could not proceed any further. Because they started to speak different languages and could not understand each other anymore.

It’s a metaphor. Metaphor that could be treated literally: why people are speaking different languages and why these differs so much. However, once we dig deeper, we can understand it’s not about the languages. It is not the language. It is the way we communicate. Our gestures, our abstracts, our perceptions. All these differs from person to person. And we can not fully align with each other and cooperate on such a level to “reach the sky”.

How it relates to AI? Well, AI is “smart” enough to understand any language, metaphors, abstractions. It had an issue with sound (music), and math. But it can “understand” abstractions. In theory, AI could be a universal translator, that unites people no matter what’s the culture, language, religion etc. Adjusting any thoughts to the level and culture of any person. Until person is physically capable of understand any level of abstraction.

But what’s the sky? The sky is a place where God lives. This Bible story tells us to avoid trying to build something that could reach (replace) God. These attempts will fail and cause long-lasting consequences. We could argue and dream how exactly this could happen. Undiscovered law of physics, a flaw in our thinking that will prevent agreeing on crucial element of system design. Or something even simpler: like greed and lack of trust to let someone unite all the resource and use it for “super-ultimate AI model”.

What could be an attempt to reach a sky? Well, here I have one specific idea in mind. With all the use cases we are currently trying to use it for. Current AI models (ML models, actually!) are well at pattern recognition. What’s the obvious move once we learn to recognize the pattern? I think you got it right. Predict the next move.

There are 2 possibilities: either ML will enable something or someone to predict the future and we’ll witness the times of antichrist, either we fail painfully.

What should we do? Let’s keep generating funny cat memes, write code, and use it to discover our own flaws and improve ourselves!