Sitting before a room full of cardinals and monsignors, at the last spot on a long table with Pope Leo in the elevated middle seat, the Canadian computer scientist and Anthropic co-founder Christopher Olah probably had a sensation of surprise. Who would have imagined his life would lead him here?
Pope Leo, in fact, when he took the microphone, specifically thanked the scientist for his presence and his presentation.
What Olah had to say in his approximately seven-minute address, one of various speakers in the presentation of Pope Leo's first encyclical, Magnifica Humanitas, was insightful.
He began by acknowledging the "incentives and constraints" AI developers face, which "sometimes conflict with doing the right thing": the pressure to stay commercially viable, to stay at the forefront of research, geopolitical pressures, and "the older, plainer pressures" of pride and ambition.
These factors will always influence, he admitted, so "if we want this technology to go well, it is enormously important that there be people outside those incentives" -- people who are "paying close attention" and "who are willing to say hard things and insist on safety."
Olah expressed gratitude to Pope Leo and the Church for taking up this work.
He said that developers have found a common conviction: "If this technology is coming, it must go well" -- well for our common home, and for the children to come.
Olah said that computer scientists are not the people to handle these questions.
How AI will interact with the world -- "how it ought to interact with the world" -- are questions for the "humanities, for religions, for philosophy, for society at large," he said.
He went on to note three questions where the "Church's voice is especially needed."
1. The duty to the global poor and how the economic gains of AI should be shared. What is to happen if AI displaces human labor at a very large scale. This is, he said, "the kind of problem the Church has historically refused to let the world ignore."
2. A moral imagination and ambition regarding human flourishing: What does flourishing actually look like? This, he said, is not a question a lab can answer, but questions that "traditions like yours have carried for millennia."
3. Finally, the discernment on the nature of AI models, and the "mystery" that he admitted sometimes even the developers discover within them. "We keep finding things that are mysterious, even unsettling," he admitted.
Olah concluded by saying that the world needs more efforts like Magnifica Humanitas. "Today is just the beginning," he said. "The start of a long collaboration between those of us who are building this, and those who can see what we, from the inside, cannot."
