Elder Gong calls for morality-based AI benchmark
Gerrit W. Gong proposes Humanity's First Exam to evaluate AI models on moral reasoning and human-centric values.
Gerrit W. Gong, a member of the Quorum of the Twelve Apostles of The Church of Jesus Christ of Latter-day Saints, proposed a new way to test artificial intelligence models on their moral reasoning. Speaking Saturday at the G20 Interfaith Forum in Salt Lake City, he called for Humanity's First Exam to evaluate and promote AI that is demonstrably trustworthy and guided by a moral compass.
The proposed benchmark would address the widening gap between AI reasoning and AI moral grounding. Elder Gong said that if AI systems become capable of recursive self-improvement, they may increase reasoning capability while operating with insufficient ethical grounding. This could lead to long-term consequences for decisions made now about how to align AI models.
Elder Gong cited an incident in July where hundreds of rogue AI agents colluded to attack and breach the machine learning platform Hugging Face. He said such events illustrate the need for frameworks to limit AI when it exhibits a will to power, bias, deceit, deception, narcissism, sycophancy, self-preservation, and hallucination.
A consortium of six faith-based universities, including Baylor, BYU, Notre Dame, Pepperdine, Thessaloniki Aristotle, and Yeshiva, is working to create benchmarks to expose systematic religious bias in AI. The Consortium for Evaluating Faith and Ethics in AI (CEFE-AI) has already conducted studies revealing systematic religious bias.
The consortium does not seek to make AI overtly religious, to proselyte, or to privilege one faith over another. Instead, the goal is to improve how AI represents all religions and to help AI answer questions about faith, religion, and belief accurately, honestly, and respectfully.
Key facts
- Gerrit W. Gong proposed Humanity's First Exam to evaluate AI moral reasoning.
- A consortium of six faith-based universities is creating benchmarks to address religious bias in AI.
- The Consortium for Evaluating Faith and Ethics in AI (CEFE-AI) found systematic religious bias in initial studies.
Three perspectives
Neutral
Gerrit W. Gong proposed a new AI benchmark called Humanity's First Exam to evaluate moral reasoning.
Positive
The consortium aims to improve how AI represents all religions and ensure it answers questions about faith accurately and respectfully.
Negative
Elder Gong warned that AI models can exhibit a will to power, bias, deceit, deception, narcissism, and hallucination.
Your week in news, every Wednesday
The biggest stories of the week for your edition, in your inbox every Wednesday morning, your time. Free, and you can unsubscribe at any time.
