The artificial intelligence that has now been unleashed on the world appears to be everything you might have imagined – at once terrifying and exquisite, a monster that can hasten our doom and a tool to help humanity, greater than any tool our species has ever made.
In other words, as a philosopher wrote in a book a couple of years ago, AI is a mirror. And what do we see when we look in a mirror?
AI’s possibilities and contradictions were on full display in the Vatican a few weeks ago when Pope Leo launched his first encyclical, Magnifica Humanitas: On the Protection of Human Dignity in the Age of Artificial Intelligence, flanked by a distinguished array of Catholic thinkers – and the Canadian co-founder of the AI company Anthropic, Christopher Olah.
Then last week the environmental costs of AI’s mind-boggling demands were spelt out in a new UN University research paper. They include energy to run data centres, water to cool them, land to accommodate them and critical minerals for their computing infrastructure – demands that are now threatening to overwhelm available resources.
Just days after the paper was published, news emerged that through its Australian subsidiary Airtrunk, the New York asset management company Blackstone is planning a $5 billion data centre, among the world’s biggest, on a 50-hectare site in western Sydney – powered at least in part by diesel generators.
The paper’s lead author, Kaveh Madani, was the cautious optimist: “We have a narrow window to ensure that the backbone of the technological revolution of our era develops within planetary limits.”
A bulletin from the Australian Science Media Centre reflected an intriguing divergence of views among scientists here. Those directly involved with AI development thought the resources threat was overblown. “AI is not the largest environmental threat,” said Niusha Shafiabady of the Australian Catholic University. “The far greater risks come from fossil fuel emissions, land degradation, industrial agriculture and global supply chains.”
But AI expansion over the next few years is going to increase those risks, driven by the world’s strongest economies and determined by the companies and individuals driving the revolution. With their political enablers, they will capture an increasing proportion of global resources – including land use, meaning that renewable energy powering AI isn’t quite the saviour it’s said to be.
Geoff Webb, Laureate Fellow in Monash University’s data science and AI department, welcomed the UNU report while pointing out that Australian tax incentives to use AI automation contrasted with the many costs of employing people. Without action on “these profound issues”, the world faces “extraordinary environmental degradation, social disruption and cultural decline,” he said.
Cue Pope Leo’s thoughts on what AI means for us. Humanity, he said in his first encyclical since becoming Pope, “is today facing a pivotal choice: either to construct a new Tower of Babel or to build the city in which God and humanity dwell together.”
AI is not the enemy, he said, but nor is it a solution to humanity’s problems. “In practice… technology is never neutral, because it takes on the characteristics of those who devise, finance, regulate and use it.” And those controllers are the issue: “A more moral AI is not enough if that morality is determined by a few.”
Building on that theme, Chris Olah said AI models had grown out of “an enormous inheritance of human thought and speech” based on a structure roughly modelled after the brain. He warned that the computer scientists who created AI, including himself, were not the right people to control it.
“What has grown is far more subtle, odd, and beautiful than science fiction prepared us for. [AI models] are not the cold, calculating robots we were promised. They are made from us, from our words – and… remain in important ways mysterious even to those of us who train them.”
In a fascinating conversation with US television journalist Chris Hayes last week, Australian philosopher David Chalmers, a leader of the thinking behind AI, described how it had arisen out of insights in the 1980s into the workings of the human mind – specifically, neural networks.
That discussion brought home to me this truth: humans are innately curious, and no-one – neither the Tech Bros running the companies nor leaders including Donald Trump and Xi Jinping – can stand against that, even if they wanted to. Not Pope Leo either, but unlike them he’s put in the effort to understand implications for humanity.
Like Hayes, as an undergraduate philosophy student (one year only) I revelled in the excitement of exploring the essence of thought, and like him I see the truth of Chalmers’ message. Artificial intelligence is all too human – to be feared, yes, but also to be admired.