Human-Centric AI
The barrier most AI programmes trip over.
"Human-centric AI" is one of those phrases that survives because it offends no one. Everybody agrees. Nothing changes.
The barrier most programmes trip over is not whether AI should serve humans — of course it should — but whose humans, and at what cost to which other humans. The customer-service AI that delights the customer often does so by removing the agent's judgement, the context the agent built over years, and eventually the agent. The "human-centric" framing hides who is at the centre and who is at the edge.
Honest human-centric design starts by naming the trade-off. If the AI improves the customer experience by de-skilling the frontline, say so. If it improves employee productivity by shifting cognitive load onto the customer, say so. The decisions are legitimate. The silence about them is not.
Our view: a human-centric AI programme is one where the humans affected — on both sides of the system — can describe what changed in their work, and whether they think it was worth it. If they cannot, the programme is not human-centric. It is AI with better marketing.
Everything else is a slogan.
The decision: name whose work the AI is reshaping, and ask them directly whether the reshape was worth it.