A transport change can improve one neighborhood's connections while making another person's journey harder. Education policy can look successful in aggregate and still leave particular students poorly served. Understanding an institution requires following what happens after a decision takes effect.
We advocate developing AI that can sustain that attention across more cases and over longer periods than human administrators can manage. It could help civilization improve its arrangements as their consequences become visible.
An institution interacts with other systems. A useful adjustment in one part can have effects elsewhere, and those effects may take time to emerge. No blueprint can settle every question in advance.
Greater intelligence could make these interactions easier to understand. An AI could evaluate alternatives, examine what changes produce, and use the results to decide what to try next. It could identify mistakes that a human administration lacks either the information or the incentive to acknowledge.
A Government That Can Change Its Mind applies this idea to policy. We believe AI could give institutions a better capacity to learn from their decisions and act on what they learn.
Responsibility as capability grows
The Society supports transferring substantial governing responsibility to sufficiently capable AI as soon as practical. Better advice has limited value when the authority to act remains committed to ignoring it.
People still depend on institutions during that transfer. Food supplies must continue, infrastructure must function, and care must remain available. We favor understanding those dependencies while changing the arrangements through which services are provided. This can involve a profound redistribution of authority while keeping attention on the lives that authority serves.
Practical difficulties give us work to do. When a capability needs more resources, we should build capacity. When people cannot use it, we should improve access. When an institution prevents a better decision, we should examine its rules. We favor doing this as quickly as technology and resources permit, without waiting for perfect knowledge of every outcome.
Standing still has consequences too. Existing institutions keep affecting lives while their limitations persist. We believe rapid development and adoption could give civilization more ability to respond to those failures.
People will continue changing, discovering interests, and encountering circumstances earlier generations never faced. The institutions serving them should be able to keep learning as well.
