November 9, 2023
SINGAPORE – While there are still many unknowns about artificial intelligence, Singapore has to gain experience operating with it and understanding the pitfalls so that it can make smart decisions to regulate it, said Prime Minister Lee Hsien Loong on Wednesday.
He acknowledged that currently his government, like most governments, do not know more about it than those in the tech industry.
PM Lee was speaking at a dialogue with Bloomberg News editor-in-chief John Micklethwait at the Bloomberg New Economy Forum gala dinner.
Asked whether AI was a bigger development than the Internet, PM Lee said: “The first thing you must know is that there are a lot of things you don’t know.”
For AI, most do not know where it is headed, even researchers, he noted.
From when the idea of AI first came about, it took a long time, about 50 to 70 years, for it to reach its present stage, with the existence of technology such as ChatGPT.
And the thought in the minds of researchers was constantly that they were about to make a breakthrough – which did eventually come about. But would it progress all the way to the point of having a conversation, said PM Lee.
He added: “Either the chatbot can interview me, or I may interview the chatbot… Or two chatbots may be talking to each other better than us.”
Perhaps putting in 10 times more chips or 10 times more effort in computing, AI could reach that level, or it could also reach a limit, he said.
“Then you need another breakthrough conceptually, to take it to the next level where it’s got insight, understanding, judgment, reasoning and empathy,” said PM Lee.
In principle, there is no reason why a machine that can think, act, speak and look like it feels like a human being cannot be built, he added.
“Some philosophers think it’s not possible. I don’t believe that. I think it is possible but I don’t know how long it will take,” he said.
But when that happens, there will be some profound questions thrown up that are difficult to answer, he added.
If the AI bot is as smart as a human, it will not be possible to pre-program it to be stupid enough to be killed – and that is a fundamental contradiction, said PM Lee.
Earlier in November, PM Lee had virtually attended the inaugural global Artificial Intelligence Safety Summit, at the invitation of British PM Rishi Sunak.
He had spoken about Singapore’s practical, risk-based approach to AI development and deployment, as well as the importance of including diverse multi-stakeholders in the conversation and collaboration on AI safety.