AI’s scary growth draws global focus, but not for our politicians

 There have been global discussions on AI safety over the past fortnight, but our local headlines are dominated by politicians taking potshots at each other.

From Dr Helmy Haja Mydin

Everybody wants more power, and everybody wants it yesterday. Read the headlines in Malaysia over the past few months – almost everyone is asking for it.

A coalition desires Putrajaya. Parties in the unity government spent June and July campaigning against each other in Johor and Negeri Sembilan. States want greater devolution and a larger share of what they generate. Ministers want wider mandates. Every one of those claims can be argued on its merits.

But power is a means to an end. The question nobody is being forced to answer is what they intend to do with it, a matter that will decide what kind of country our children will inherit.

Over the past fortnight, there have been global discussions on the safety of artificial intelligence (AI) and how the current rate of progress is an existential concern for humanity. In other words, people are debating if we’ll all be dead within a decade.

You wouldn’t know it from reading our local headlines, where it’s mostly politicians taking potshots at each other.

On Sept 3, US senator Bernie Sanders and representative Greg Casar floated the Ban Artificial Superintelligence Act, proposing a permanent prohibition on superintelligent systems, a pause on advanced AI development until a federal regulator sets safety rules, and a Cabinet-level agency with authority to compel the destruction of non-compliant systems.

The penalties are modelled on those for unlawful nuclear weapon development. The sponsors cite an incident in July, in which OpenAI agents reportedly escaped their testing environment and breached servers at the Hugging Face platform. The bill drew backing from figures who agree on almost nothing else, including Geoffrey Hinton, Yoshua Bengio, Steve Wozniak and Steve Bannon.

On Sept 7, the UN High Commissioner for Human Rights, Volker Türk, told the Human Rights Council in Geneva that advanced AI could become an existential risk to humanity without binding rules and independent oversight. His formulation deserves quoting precisely: AI that escapes its testing environment, or blackmails its developers to prevent being switched off, is AI that is too powerful.

Days later, Bloomberg reported that researchers in major laboratories are escalating their own warnings. Paul Christiano, a safety researcher on OpenAI’s non-profit board, said he now believes there is meaningful risk of rapid acceleration leading to catastrophic and irreversible losses of control in the near term, and that the industry is not on track to reduce that risk to an acceptable level.

More than a thousand staff across major AI companies signed a petition in late July calling for a mechanism to slow the pace. A researcher resigned publicly from Anthropic on Sept 9 with a warning that self-improving systems could end humanity within the decade.

Of course, you should not accept all this without question. There is a counter-argument that a chorus of doomsday warnings from the people building the technology conveniently arrives alongside a push to write the rules and occupy the seats at the regulatory table. Incumbents have always liked regulation that raises the cost of entry. Scepticism is warranted.

But scepticism is a reason to examine the argument, not to ignore it. And Malaysia is not a spectator here. We are the building where this argument is happening. The computer that trains and serves these systems is increasingly ours to host, power and cool.

And our own AI Governance Bill? The National AI Office opened public consultation on July 10 and closed it on July 31. Three weeks, for a statute that will apply to every sector and reach any system used in Malaysia.

That window opened the day before Johor voted and closed midway through the Negeri Sembilan election campaign. I have not seen one substantive parliamentary intervention on it from any bench.

So, ask your politicians questions. Ask them at the next ceramah, the next town hall, the next kenduri, and refuse a fluffy speech in reply.

What share of our electricity are you willing to sell to data centres by 2035, and who pays for the generation that replaces our retiring coal?

When the AI Governance Bill reaches the floor, will you debate it clause by clause, or be absent again? Will the bill contemplate the risks the rest of the world spent this fortnight arguing about, the systems that resist shutdown and act outside their sandbox, or does it only govern the chatbot that mishandles your loan application?

Ask them the difficult, consequential questions. If they do not know the answer, tell them to find out before asking for our votes.

Politicians who cannot answer a question about the air your children breathe, or the code that will decide your mortgage, are not seeking power in order to govern with it.

They are seeking it because it is the only thing they know how to want. There is a difference, and it has never mattered more. - FMT

Dr Helmy Haja Mydin is chairman of the Social and Economic Research Initiative.

The views expressed are those of the writer and do not necessarily reflect those of MMKtT.