ChatGPT is not an ‘intelligence’

I found this Short on Youtube just now, and it highlights just how very not intelligent AI is. In this example, it’s ChatGPT but the principle behind it is the same for the others:

Statistically, there are far more pictures of watches showing 10:10 than there are of watches showing other times. Ditto for pictures of hands writing: they are almost all of the right hand, not the left. And these Large Language Models [LLMs] use the values provided by statistics – from the internet – for their answers to questions. But statistics, especially on the internet, can be horribly skewed, sometimes by accident, often by malicious intent.

Let me repeat that, statistics on the internet can be horribly skewed by malicious intent. So how will we ever be able to trust the hallucinations provided by so-called ‘AI’?

Meeks