Manj Chenna

The vision · Capability

Capability is downstream

The models keep getting better, and that is the story everyone tells. It is the wrong story to obsess over. Capability is an input. The rules that decide what it is used for, and who answers when it is wrong, are the output that actually matters.

scroll to read
Manj Chenna · Founder, Sanctity · Building real human oversight of AI · Amsterdam

Capability is downstream of the rules. A more capable model does not choose its own purpose, its limits, or who is accountable for it. Those are set by people, in the design and the governance, and they decide where all that capability points. This is why I think the more important contest is not how powerful AI gets but who sets the rules AI runs on, and why real human oversight of AI is the part I chose to build.

a world these systems make bearablea world they make efficient andunanswerablethe rules underneathone lever, two futuresA SMALL MOVE AT THE BASE, A LARGE ONE AT THE FAR END
ONE LEVER AT THE BASE: THE RULES UNDERNEATHa world these systems makebearableWHERE IT COULD POINTa world they make efficientand unanswerableWHERE IT IS POINTINGA SMALL MOVE AT THE BASE, A LARGE ONE AT THE FAR END
The curve is going up either way. The argument worth having is about the lever at its base, which almost nobody is allowed to touch.

Why capability gets the attention

Because it is visible and it is measurable. A benchmark goes up, a demo gets better, and progress feels concrete. The rules are quieter and harder to measure, so they get less attention, even though a small change in them redirects the entire capability behind them. Watching only capability is like watching the engine and ignoring the steering.

Why the rules matter more

Because the same capability can serve almost any end, and the rules are what choose. Two companies with the identical model can build one system that keeps a human in command and one that does not. The difference is not the capability. It is the governance, and the governance is the part you can still decide.

Read on

See the cornerstone, who sets the rules AI runs on, and governance is a product.