Artificial intelligence has the potential to outperform traditional methods in predicting the movements of tropical cyclones, the Observatory's director said on Saturday.

Chan Pak-wai said the forecaster is seeking funding on the use of AI, hoping to include information exclusive to the territory in the model.

But the head of the Observatory conceded that the technology is still at an early stage in forecasting weather.

"Over the next years, there are going to be more further improvements on the model, so hopefully it can beat the traditional method of forecasting the movement of tropical cyclones - maybe in the future. But right now, I can only say that it gives you an additional piece of information to consider," he said after a radio programme.

"In a tropical cyclone's forecasting or warning, one particular important aspect is to know beforehand what the possible scenarios [are] for the future movement of tropical cyclones, so that you get better prepared. In that aspect, AI is really providing a useful reference."

Chan said it's been difficult to track Super Typhoon Doksuri, which was closest to Hong Kong a week ago, as well as the current Typhoon Khanun.

He pointed out that global warming would lead to a higher proportion of severe typhoons, slow down the pace of tropical cyclones and change their paths, with many of them now heading north to South Korea and Japan.

He also attributed global warming to more frequent extreme weather events, while calling on the public to do their part in reducing carbon emissions.

"I need to turn on air conditioning when I am at work. Because we are located on a hill and my office faces west, so it gets very hot during the day. But I never turn on the air conditioner when I'm sleeping at night," Chan said.

"On using electricity and going plastic-free, we should do our best and do as much as we can."