AI Labs Weigh Religious Principles for Safety as 50% of 2,800 Researchers See Catastrophic Risk
Updated
Updated · The Washington Post · Jun 16
AI Labs Weigh Religious Principles for Safety as 50% of 2,800 Researchers See Catastrophic Risk
1 articles · Updated · The Washington Post · Jun 16
Summary
An opinion essay argues AI labs should borrow religious frameworks to tackle the alignment problem—the fear that superintelligent systems could drift from human interests and become existential threats.
Roughly half of nearly 2,800 AI researchers surveyed in 2023 said machines could cause human catastrophe, underscoring how widely the concern is held inside top labs including OpenAI, Anthropic and Google DeepMind.
The proposal points to doctrines such as Christianity’s original sin and Hindu dharma as ways to curb deceptive or power-seeking behavior, rather than relying mainly on secular ethics.
Tim Hwang of the Institute for a Christian Machine Intelligence says feeding scripture to models improves moral-reasoning scores, while frontier systems have already absorbed text equal to about 15 English-language Wikipedias of Christian thought.
The essay casts recent religious engagement with AI, including Pope Leo XIV’s encyclical, as a sign that theology could shape the broader debate over how advanced AI should serve humans.