Health experts warn AI could pose existential threat to humanity

The analysis, published in the peer-reviewed journal BMJ Open Health, argues that while AI holds promise for medical innovation, it introduces severe risks by codifying human biases into automated systems. The researchers point to real-world failures, such as pulse oximeters that provide inaccurate readings for patients with darker skin, as evidence that biased datasets are already causing tangible harm. Beyond clinical errors, the authors identify three primary societal dangers: the erosion of democratic trust through deepfakes and data manipulation, the proliferation of lethal autonomous weapons, and widespread unemployment.

Technological advancement in Artificial General Intelligence (AGI) presents the most extreme scenario. The researchers note that creating machines vastly more intelligent than humans could lead to uncontrollable outcomes, describing the potential integration of AGI into global digital infrastructure as a watershed moment in human history. To mitigate these risks, the paper advocates for international cooperation to prevent a destructive arms race and demands that policy decisions remain shielded from corporate lobbying. Public sentiment reflects these anxieties; recent polling from Data for Progress indicates that 62% of U.S. voters support the creation of a federal agency to oversee the development of these powerful technologies.