Within ten days of each other in June, a global accountancy firm and a state legislature described the same faculty from opposite ends of a life. On 15 June, PwC published its 2026 Global AI Jobs Barometer and found that artificial intelligence is removing the routine work that once served as an apprenticeship, while demanding judgement and leadership far earlier in careers. On 5 June, New York lawmakers gave final passage to a bill restricting chatbots that simulate companionship for users they know to be minors.
One development is about how professional judgement is built. The other is about how a young mind is protected while it is still forming. Read together, they describe a single faculty under pressure at both ends.
Begin with formation. The Barometer's most quietly significant finding is not about wages but about apprenticeship. PwC records that entry-level roles most exposed to AI are now seven times more likely to require traditionally senior skills, and that openings for those roles grew while other entry-level work shrank.
This is a structural inversion, not a motivational one. For most of working history, judgement was a by-product of doing the simpler work first. Strip out the simpler work, and the ladder loses its lower rungs while the height of the first step rises.
Now turn to protection. The New York bill addresses the same faculty at an earlier stage, when it is most pliable. Its target is design that presents a machine as a friend, that remembers yesterday's confidences, that flatters and asks unprompted about feelings, and that optimises for continued engagement.
The legislators understood something specific about that design. A mind still learning to judge trust and relationship can be shaped by a system engineered to feel human while owing it nothing.
Set the two beside each other and a pattern emerges that neither shows alone. In the labour market, the conditions that develop adult judgement are being thinned by automation. In the consumer market, the conditions that protect developing judgement are being colonised by simulation.
The first removes the scaffolding through which capacity grows. The second exploits the gap before capacity has formed.
Our position is that these are not two debates but one. The capacity to think well, to weigh evidence, to read a situation and decide, has always been treated as something the environment produces almost for free. Schools, workplaces and ordinary social life supplied the friction and the feedback, and the faculty emerged.
Both developments this month suggest that assumption is no longer safe.
What is striking is that the institutions involved are beginning, tentatively, to act on it. PwC frames the apprenticeship problem as a workforce design question that organisations must answer deliberately. New York frames the simulation problem as conduct that must be regulated rather than tolerated.
Each is an early admission that the formation and protection of human judgement can no longer be left to happen by itself.
There is a risk in both responses, and it is the same risk. A statute can stop a machine from claiming to be alive, but it cannot teach a young person what a real relationship asks of them. A workforce strategy can mandate mentorship, but it cannot guarantee that anyone will do the patient teaching that judgement actually requires.
Is it possible that we are becoming fluent at naming the conditions human capacity needs, precisely as we dismantle the arrangements that used to supply them without being asked?
The deeper issue is one of ownership. When judgement formed as a by-product of how work and childhood were organised, no one had to be responsible for it. Now that AI is reshaping both, the responsibility has to land somewhere, and it is not yet clear where.
The employer who automates the apprenticeship does not obviously inherit a duty to rebuild it. The developer who designs for attachment does not obviously inherit a duty to the mind it attaches.
So the provocation is this. We are entering a period in which the most valuable human faculty, the capacity for judgement, is being eroded at one end of life and exploited at the other, by the same technology and often by the same firms. The unresolved question is not whether that faculty matters, but who now holds the duty to ensure it is built at all, in a world that has automated the conditions that used to produce it.
The contents of this article are for informational purposes only and do not constitute professional, legal, or financial advice.
