A new preprint on arXiv examines the dangers of an AI race. The abstract notes that many estimates place the emergence of artificial general intelligence within the next few decades, possibly sooner. Once AGI arrives, it may rapidly exceed human intelligence, potentially by a wide margin.

The paper's title refers to a 'triple risk' associated with an AI race, but the abstract does not spell out what those three risks are. Readers would need to consult the full text for specifics. As a preprint, the work has not yet undergone formal peer review.

Despite the limited detail in the abstract, the framing highlights a growing concern among researchers: that competitive pressure to build AGI could amplify existing dangers rather than simply accelerate progress.