Taiwan’s first Digital Minister made the case that democratic deliberation isn’t too slow for the AI era – it may be the only thing fast enough to keep up. 

The conventional wisdom about government and technology goes like this: democracies are too slow, too cautious, and too consensus-bound to keep pace with the speed of artificial intelligence. To compete, they must streamline, centralize, and act. 

On the Main Stage at the GovTech 4 Impact World Congress in Madrid, Audrey Tang spent the better part of an hour dismantling that assumption. 

In a wide-ranging fireside chat with Laura Gilbert, Senior Director of AI at the Tony Blair Institute for Global Change, Audrey argued that the real threat to public institutions is not that they move too slowly, but that they have stopped listening at the scale modern problems demand. The conversation ranged across democratic innovation, misinformation, deepfake fraud, and the responsible use of AI, but it kept returning to a single question: how can a government move quickly without losing the trust that makes it legitimate in the first place? 

Tang’s answer turned on a phrase she calls “plurality” — designing digital systems that surface the unexpected overlap between opposing camps rather than the rage that drives engagement. She offered a striking example from Taiwan’s response to a wave of deepfake financial scams, where structured public deliberation produced real legislation, fast, and with broad public support. Deliberation, so often dismissed as the enemy of speed, had delivered both.

And she left the room with a warning that lingered. Send a robot to the gym in your place, she suggested, and the scores may look impressive, but your body still weakens. The danger of AI is not only that it makes mistakes. It is that it may tempt us to stop practicing judgment altogether.

What she meant by that and how Taiwan is building systems to prevent it, is worth hearing in her own words.

Watch the full conversation

The full fireside chat covers far more than we can capture here: Tang and Gilbert on misinformation, algorithmic bias, the difference between “human in the loop of AI” and “AI in the loop of humanity,” and how other governments have adapted Taiwan’s methods, for better and worse. Watch the Full Session: