[ExI] Claude for president?

[Adrian Tymes]

On Mon, Mar 16, 2026, 5:42 AM Jason Resch via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

> I see this as a practical problem, and one that could be engineered around
> if obtaining repeatable output were the goal.
>

If it was the primary goal, overriding any other goal, then yes I agree
this would be possible.  The problem is that it is practically impossible
to make this the primary goal in the scenario presented.

In order to be selected for the role, advisors to the President of the
United States must be high powered, in whatever sort of "power" the
President cares about.  (This does not necessarily equal "competent", as
demonstrated by several members of the current and previous
administrations.)  In the case of AI advisors, replicability of advice
seems highly unlikely to factor in among the types of power that will be of
primary value to the President.  It is not inherently undesirable in and of
itself, but it necessarily falls by the wayside compared to the qualities
that will be selected for.  I use the term "power" in this description to
make it obvious what qualities of an AI advisor would be selected for
instead.

This is inherent to their function. and can not simply be legislated away.
Any attempt to over-prioritize a lesser goal (such as replicability) by law
will simply fail in practice.  Furthermore, the probability of that failure
is so obviously high (specifically, obvious to Congress, as each member can
readily imagine themselves in the President's office) that it seems
improbable that Congress would ever try passing such a law.

>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20260316/0e772f8d/attachment.htm>