Human-Centered AI (HCAI)
What is Human-Centered AI (HCAI)? — and how HCI, HAI and HCAI differ
What exactly does it mean to build AI around people?
Human-Centered AI (HCAI) is the commitment to design AI so that it enlarges human judgement and control rather than replacing it. It refuses the trade-off between automation and human control and aims to raise both, and it asks at design time who pays when the system is wrong.
HCI, HAI, HCAI — one sentence each
HCI (Human–Computer Interaction) is a field: everything that happens between people and computing systems. HAI (Human–AI Interaction) is a subfield of it, narrowed to AI. HCAI (Human-Centered AI) is not a field but a stance — it settles not what you study but what you are building toward.
So the three do not compete. You work inside HCI, study HAI, and apply what you find against HCAI as the standard. That order holds.
What human-centered actually rules out
A phrase that rules nothing out is worth nothing. In practice this standard does exclude things: designs a user cannot leave, automation that never hands control back, opacity that keeps a person from even knowing the system was wrong, and the kind of success where the metric climbs while the person's life gets worse.
Why I hold it as a direction
HCAI is less a state you can declare reached than a question you have to ask again each time. Running a product, the cheapest way to move a metric and the better thing for the person keep diverging. Deciding in advance which one you take is what this stance actually amounts to.