[ExI] Douglas Hofstadter, LLMs and self-referentiality

[Adrian Tymes]

On Wed, Sep 2, 2026 at 2:32 PM Mike Dougherty via extropy-chat
<extropy-chat at lists.extropy.org> wrote:
> Judging from average observed behavior, i am unsure if any of us are human
>
> Spike has already established that in his misspent youth he hosted an unsecured digital impersonation party for the purpose of plausible deniability of his own digital signature (that running joke keeps getting funnier)
>
> I'm unsure if my own use of AI is any more secure than that.
>
> So no, we're all connected at this point.  Resistance is futile, we are Borg

I claim not to be an AI myself, even if I sometimes speak to them, but
if we're confessing stuff like that...I may have already shared this
tale, but it's relevant.

Back in the mid-90s when I was at college, I sometimes posted to
Usenet from my Computer Science account.  At one point, I found a
thread where someone was claiming to be me, but saying things I
disagreed with.  A quick inspection showed "Adrian Tymes" posting from
somewhere in the humanities.  So, I joined the thread as "Adrian
Tymes" from a dorm-linked account and "Adrian Tymes" from my
CompuServe account, all three of the "me"s expressing different points
of view alongside the fake "me".  Eventually, whoever it was gave up
and stopped posting as me in the groups I was watching (and, so far as
I know, never impersonated me anywhere else).  Nothing approaching the
level of what we'd now consider to be a digital signature was
involved; they apparently simply put my name in their "unverified
human name to attach to posts from this account" field.