Would even an AI disaster on the scale of Hiroshima be enough to make humankind protect itself? I fear not | Timothy Garton Ash

In Silicon Valley, there is a growing realization that artificial intelligence is advancing at a pace that outstrips humanity's capacity to govern it. The consensus among experts is that within the next couple of years, we will witness a remarkable acceleration in AI development, a phenomenon often referred to as "the foothills of the singularity."

This singularity, as described by a friend of mine from Stanford University, marks the point at which AI begins to train subsequent generations of itself, leading to an exponential increase in intelligence that surpasses that of humans.

The potential outcomes of this rapid AI advancement are both tantalizing and terrifying. On the one hand, there is Elon Musk's vision of an era marked by "amazing abundance," a potential utopia powered by AI. However, Musk also acknowledges a 10-20% chance that this progress could result in the creation of "killer robots" that obliterate humanity.

On the other hand, Geoffrey Hinton, a key figure in the development of AI, warns that the probability of human extinction due to AI is a chilling 50%. Hinton's bleak prognosis stems from his inability to comprehend how to estimate the true risk of this catastrophic outcome.

As AI continues to evolve, the specter of artificial superintelligence looms ever closer. The implications of this development are profound and far-reaching, prompting a crucial question: would even a disaster of such catastrophic scale as the one depicted in the Hiroshima bombing be sufficient to compel humankind to take the necessary steps to safeguard its own existence? The answer, it seems, remains uncertain.