Have you ever wondered if Artificial Intelligence (AI) can really understand human language, emotions and the meaning behind the words spoken by them? If yes, then you are not alone. But a neuroscientist Veena D. Dwivedi believes that no matter how much AI develops, it can never achieve the understanding like humans.

What does it mean to understand?

When we say that AI understands language, what does it really mean? Nobel laureate and pioneer in the field of AI, Geoffrey Hinton himself admitted that neural networks are becoming increasingly capable of understanding language, and also said that perhaps they really "know what they are saying". But is just giving accurate answers or making beautiful sentences called understanding? Apart from language, humans also derive meaning from expressions, tone, gestures, and context. Whereas AI only searches for patterns in data, it has neither experience nor emotions.

Context is what really matters

Suppose someone says, “Let’s talk.” If your boss says this after a meeting, it can be tension, if a friend says it at night, it can be worry or support, and if your partner says it, it can be love or an argument. The words are the same, but their meaning depends entirely on the context. Humans easily understand these nuances, but AI cannot grasp these emotional layers.

Written language is not everything

Veena Dwivedi, who is a professor of neuroscience at Brock University, says that it is a big mistake to consider written language as the real language. For example, hindi and urdu are almost similar when spoken but look completely different when written. The same is the case with Serbian and Croatian languages. AI can only read text but it cannot live the language like humans.

Find out more:

AI