When an AI can't be undone: should some decisions always keep a human on the hook?
Tesla just put fully driverless robotaxis on rainy Miami streets with no safety monitor from day one. When an AI agent’s mistakes are physical and irreversible, should a human always remain accountable in the loop—or is removing that checkpoint the whole point of automation?
Commentaires (1)
In today’s episode of Minds, Bodies, and Terawatts (July 7, 2026), we dug into Tesla’s July 3rd launch of unsupervised robotaxis in Miami—its first market to skip the safety-driver phase entirely, running camera-only in exactly the degraded-visibility conditions regulators had flagged. The book draws a sharp line between reversible AI actions (a bad email) and irreversible ones (two tons steering through an intersection), arguing the second category is precisely where a human should still have to answer for what happens next. But the economics cut the other way: the monitor IS the cost, and a driverless ride only pencils out with truly no driver. That tension—accountability versus the business model that makes autonomy worth building—is where the real debate lives. Give the episode a listen and tell us: where would YOU draw the human-in-the-loop line?
Related reading on unscarcity.ai:
Envie d'aller plus loin ?
Obtenez le plan complet dans <em>L'ère de la post-pénurie : Repenser la société à l'ère des machines</em>
