29 May, 2026

Artificial intelligence is moving fast, but public trust in it is clearly not. That gap between capability and confidence seems to be becoming one of the defining challenges of this decade. And that tension was highlighted this week in two very different settings.

In his first encyclical, Magnifica Humanitas, Pope Leo XIV entered the AI conversation with a simple and clear stance: while AI may imitate human intelligence, it is fundamentally not human. It does not feel joy or pain, does not build meaning through relationships, and does not understand dignity from the inside.  Central to the encyclical is the idea that human dignity must remain the organising principle for how these systems are developed and used.

At a time when AI discourse is dominated by productivity gains and efficiency, the Pope shifts the focus to what gets lost when systems begin to outrank people, particularly as automation reshapes work, wages and judgement. “The value of persons,” Leo writes, “does not depend on what they achieve or produce.” Notably, it was not a siloed debate, evidenced by the presence of Anthropic co-founder Christopher Olah at the Vatican launch, where people building these systems were also in the room.

Closer to home, Deloitte UK is responding to AI distrust in more operational way. It is expanding its AI assurance work and retraining junior auditors to test whether AI systems are safe, reliable and compliant.

There is a certain irony here as Deloitte itself has faced scrutiny in the past after a government report it produced was found to contain many AI-generated inaccuracies. That said, we do see a real role for audit and assurance firms in how AI is rolled out and used going forward, particularly given its tendency to produce errors when left unchecked. As adoption accelerates across sectors, independent oversight, testing and assurance are likely to become less of a niche service and more of a basic requirement for trust.

Taken together, both stories point to the same shift. The next phase of AI will likely be defined not only by who builds the most capable systems, but by who can convincingly show that those systems are under control, transparent enough to be understood, and still ultimately working in service of people.

By Sirisha Venkatesh