There needs to be a ‘killswitch’ to avoid AI going overboard – AI Expert

There needs to be a ‘killswitch’ to avoid AI going overboard – AI Expert

KUALA LUMPUR: A form of regulation to oblige every artificial intelligence (AI) developer to complete a program with a proper kill switch should be created as a safety measure.

AI expert, Dr Poo Kuan Hoong, said that although the technology’s capabilities are not yet very advanced, it is developing rapidly and the move is necessary before it potentially threatens humans.

Currently, generally all AI that is developed is focused on doing a specific task, called “Artificial Narrow Intelligence” (ANI).

But moving forward, he said, it is certain that the potential of AI wants to be developed to a level called “Artificial General Intelligence” (AGI) which has wider capabilities, and this needs to be given more attention.

“People need to realize that AI is developing very fast and sooner or later we need to have something like a national committee to regulate all companies developing AI.

“They need to develop a killswitch to turn it off if a system or machine becomes more ‘self-aware’ and may do something unethical,” he said in the Three Corners Dialogue program on Thursday.

Kuan Hoong said again, the regulation also needs to establish that the potential of AI has limits, such as not being able to modify human genetics and only being limited to the genetics of plants and animals for example.

Share This


Wordpress (0)
Disqus (0 )