[maxsanic]

Let's face it. In any specialization, about 70% of the trained personnel are of 'normal' levels. Out of the 30% or even less who are experts, many are researchers (these people think of new ways of fine-tuning algorithms). 

In Singapore, the shortage is largely in the 'normal' category, since they mostly make use of tools that are already developed. Neither do these people develop new algorithms, or at least their focus isn't on this. 

It's not just my own opinion. I've spoken to many AT specialist and CTOs who have the same thinking. They explicitly told me that AI engineer can be trained in a matter of months. How these people can get 6k and above is attributed to hype. Even the engineers who haven't dealt with AI before succumbed to the supposed difficulty simply because the big name of 'AI'. In truth, after they work on it, they'll realize AI is not that complicated.

Ok, but this issue has been around for some years already. If like you say the CTOs also think it can be easily done by spending a few months to train much cheaper engineers, why isn't everyone doing it?

Why is the starting salary for such jobs still increasing exponentially? I can accept maybe first 3 months situation not clear companies kalan kabo so overpay for talent, but AI talent shortage has been around for at least 3 years and till now the shortage is getting even more acute, salaries even higher. Why aren't companies doing this 3 month engineer conversion on a mass scale? Honestly I find it hard to attribute this to just "government and Singaporeans mentality being inflexible".