[ExI] Against Against AI Handover

[Adrian Tymes]

On Mon, Oct 26, 2026 at 6:45 AM John Clark via extropy-chat
<extropy-chat at lists.extropy.org> wrote:
> The minimum viable change is a constitutional amendment allowing an AI to run for president. Suppose that 2028 election has five candidates: JD Vance, Marco Rubio, Gavin Newsom, Alexandria Ocasio-Cortez, and a perfectly-aligned Claude 10. And suppose that in every measurable, empirical way, Claude 10 is better.

That is an impossible premise.  "Better", in this case, is in the
subjective opinion of voters - who famously do not universally agree
on what is better.  Some would in fact prefer - as expressed by who
they would vote for - a dumber President just for being dumber, so a
smarter candidate would be a negative in their opinion.  Some will, of
course, prefer a smarter President.

A smaller but more immediately plausible step toward this might be for
some Green or Libertarian candidate to take extensive AI advice for
policy on issues where they or their party do not have strong
opinions.  What is the Green Party's view on abortion, or on sending
armed assistance to Ukraine?  What is the Libertarian Party's view on
issues that are inherently collective, and "everyone just do their own
thing" has nearly always (at least in recent history) resulted in
harms to the common and poor majority (who at least know enough to
vote against someone saying "everyone just do their own thing" on
these issues), such as water quality and air quality?

More concretely measurably, use AI-run fundraising to raise at least
$100M for the presidential cycle (at an average $100 individual
contribution, this means getting at least 1M people to contribute),
then AI-managed advertising campaigns (focusing more on where to spend
than on the creative content) to get people to vote for them and their
candidates.

Further and perhaps most importantly, recruit and run (with further AI
assistance if necessary) down-ballot candidates in at least a majority
of House, Senate, and state governor races (among those seats up for
election in that cycle).

> If an AI aligned to your values that was smarter than you were, and had devoted much more time to thinking about the question, told you that your disagreement was wrong, and actually your values would be better served by adopting this system, would you believe it?

If it truly was smarter than I to the degree suggested, and was truly
aligned to my values, it would know how to convince me and this would
be a nonissue.  Therefore, that this is an issue worth considering
suggests that the AI in question would either not be smarter than me
or not aligned to my values, despite other parties claiming it would
be both.