Geoff Mercer

A siren call is defined as a powerful temptation that’s hard to resist – but which may have negative consequences.

In Greek mythology, sirens were creatures who sang so beautifully that sailors were drawn irresistibly toward them, steering their ships into rocks and dying as a result.

I’ve come to realise the lure of artificial intelligence probably qualifies as a siren call; I’m a user and I’m feeling the pull.

Notwithstanding what was once the stuff of science fiction – AI going rogue and rendering humans subservient – the immediate negative consequence we need to focus on is competition for electricity, which fuels the computational power needed for AI to operate.

In March this year, Datagrid NZ obtained resource consent to build its hyperscale data centre at Makarewa, Southland.

Publicity at the time indicates the facility will consume up to 6 percent of New Zealand’s electricity – second only to the Bluff aluminium smelter, which uses 13 percent.

The electricity for Datagrid NZ will come from gentailer Mercury Energy, which has bought a 12.7 percent stake in the company for $53 million.

Obviously, the emergence of even more data centres in New Zealand will lead to higher electricity prices. Increased competition for a commodity hikes prices.

Unless some mechanism emerges that can pause construction of additional data centres, more will spring up.

New Zealand is seen as an ideal location because of its maritime climate, which reduces data centre cooling costs, our country’s political stability (although, that’s debatable while we’re in the silly-season run up to the election), and its cheap and largely renewable energy supply.

The Eastern Bay, with its ability to produce renewable geothermal and solar energy, is perhaps a candidate location already.

We need to have a national conversation about the wisdom of apportioning more of our electricity supply to data centres, which provide few jobs beyond their construction.

They will inevitably lead to more expensive electricity for homes and businesses.

So, to be clear, when your AI engine next serves you up information in moments, ask yourself if you’ve fully considered the economic and social impact of the sugar-rush hit you’ve just experienced.

We need to seriously consider the harms – the negative consequences of AI investment – to avoid striking rocks in the future.