Is AI Intelligent?

Can we build machines that “think”? A simple question that led to the idea of AI in the 1950s.

AI can be described as the effort by machines to simulate human intelligence - our ability to learn, reason, understand language, and so on.

Machines learn from human input (big data) and, in essence, produce what we already know. But they are incapable of originating anything…yet. That is the next step: Artificial General Intelligence(AGI), which is still out of reach.

Should we compare AI with human intelligence?

How do you test the “intelligence” of a machine? Asking whether AI is as intelligent as we are assumes the two can be weighed on the same scale.

AI as we know it today is best described as “cognitive automation”; it simulates human intelligence by encoding and applying human skills and knowledge. It excels in solving domain-specific problems where there are enough examples to learn from.

What humans have is “cognitive autonomy”, which is a different beast. Cognitive autonomy is the ability to adapt to unfamiliar problems and learn from them.

The key difference is adaptability.

Today’s AI struggles when a problem falls outside what they were trained on. That’s why AGI is still out of reach, as it requires training for every possible scenario.

So even with advancements in technology, AI will remain a powerful tool and not a sentient being.