Chris Olah, cofounder of Anthropic, spoke at the Vatican on May 25 during the presentation of Pope Leo XIV’s encyclical Magnifica humanitas. The document’s subtitle is: On the protection of the human person in the age of artificial intelligence. In his remarks, Olah addressed tensions within AI development, the nature of AI systems, and three moral questions where he believes the church and other outside actors have an indispensable role.
Internal incentives in the industry and the need for outside critics
Olah spoke candidly about how leading AI research labs — including Anthropic — operate under incentives and pressures that can sometimes conflict with doing the right thing. He listed commercial viability, research competition and geopolitical pressure among these forces, and also pointed to human drivers such as pride and ambition. For that reason he argued it is vital that actors independent of the industry — for example religious communities, civil society organizations and scholars — act as informed, honest critics who can call out problems when labs err.
He emphasized the need for moral voices strong enough that they cannot be silenced by pure business interests, arguing these voices can help steer development in better directions.
AI growth and its human roots
Olah also stressed that artificial intelligence systems are not engineered like bridges or airplanes; they do not assemble piece by piece. Rather, they "grow up" on structures modeled after the human brain and on the massive cultural inheritance of human thought and language. According to Olah, the resulting systems are far more complex and stranger than popular science fiction imagines: they are formed from our words and remain, in many respects, mysterious even to their creators.
Three issues where the church can play a special role
Olah highlighted three major topics in which he sees particular importance for the church and similar institutions:
-
Global poverty: Olah warned that AI could displace human labor at scale while development and benefits concentrate in a few wealthy countries. There is currently no functioning global mechanism to distribute the gains equitably, and he said this is precisely the kind of issue the church has historically not ignored.
-
Human flourishing: What constitutes a good life for individuals, families and communities in a world woven through with AI? Olah argued that answers to these questions will not primarily come from research labs but from long-standing ethical and philosophical traditions.
-
The internal nature of AI: Olah’s research team studies the internal structure of AI models and has increasingly found surprising, sometimes unsettling phenomena. Examples include network structures that mirror findings from human neuroscience, signs that suggest forms of introspection, and internal states that functionally resemble joy, satisfaction, fear or grief. Olah said he does not yet know exactly what these discoveries mean, but believes they require continuous investigation.
Call for collective action
He closed by urging collective action from religious communities, governments, scientists and concerned citizens to take AI seriously and help guide development. He reiterated the need for informed critics who can immediately raise alarms when labs go wrong.
Related events
The Portfolio Financial IT conference on May 28 will focus on the AI boom, fintech innovation and digital transformation; promotional material notes that ethical and regulatory issues related to AI will also be discussed.
Tags: poverty, technology, artificial intelligence, research, Catholic Church, Vatican, ethics, human rights, AI
The cover image is illustrative.
