Re: Friendliness SOLVED!

From: Edward Miller (
Date: Wed Mar 12 2008 - 16:36:14 MDT

Lay off the drugs.

You are just talking about rational choice theory,
which neither says much of anything about human action
which is frequently irrational, nor does it ensure
that an AGI would choose to be friendly.

At the very least you could have tried to prove this
with symbolic logic or something. If you had tried,
I'm sure you would find the formulas don't add up.

Take one of Eliezer's examples of an AGI that loves
smiley faces and ends up tiling the universe with
smiley faces. How would any of this prevent that? It
wouldn't. Acid test over, you lose.

--- wrote:

> Yes, I actually *AM* that arrogant -- but this
> mailing list is the ultimate acid test and I think
> that I can survive the ridicule and ignominy if I'm
> totally/horribly wrong.
> I am declaring Crocker's Rules on myself with the
> "waste of my time" exception (i.e. I reserve the
> right to be rude to anyone who is rude *AND* wastes
> my time with unusably incomplete, incoherent,
> irrelevant/spurious and/or totally repetitive
> arguments :-).
> All concise, coherent, thoughtful requests for
> clarification are welcome and concise, coherent,
> thoughtful points for debate are eagerly requested.
> = = = = = = = = = =
> Executive Summary
> I've been arguing on the AGI list that a great way
> to ensure the stability of Friendliness for AGI
> would be to make it so that it was in an entity's
> own self-interest to be Friendly (I've been calling
> this the Attractor Theory of Friendliness).
> It's obvious from hindsight, but while working on
> it, I had the epiphany that it was only to my own
> benefit to declare Friendliness myself (duh, if it's
> in any AGI/entity's self-interest, then it must be
> in *my* self-interest).
> I have therefore declared Friendliness (codifying it
> in the declaration below) and am attempting to
> spread it since it can be used to derive useful
> ethical guidelines for AGIs, humans, and all other
> entities to follow because it is in their own best
> self-interest <and mine> to do so).
> I am also promoting my catchy new slogan
> "Friendliness: The Ice-9 of Ethics and the Ultimate
> in Self-Interest" as a good way to spread
> Friendliness. :-)
> = = = = = = = = = =
> A Little More Detail
> I believe that I have come up with a formulation of
> Friendliness that is stable under self-improvement
> because, *as far as I can tell* (the big gotcha), it
> is in an entity's own best self-interest to be
> Friendly (according to the formulation -- but with
> the realization that there *probably* is a
> super-Friendliness that is even better that is a
> subset of this Friendliness).
> This formulation is simple enough that the
> intelligence required to analyze whether an action
> is Friendly or not (according to the formulation) is
> substantially below the intelligence required to be
> a significant threat to humanity.
> Absolute proof that the formulation will always
> function as we desire under all conditions is beyond
> my capabilities (at this point in time) but since
> the derivable conclusions of this formulation
> include many of the generally-agreed but
> previously-difficult-to-prove points of ethical
> behavior and do not contradict any points of ethical
> behavior that I know of, I believe that we should be
> as safe as is ethically possible (i.e. if it comes
> down to a choice between humanity or ten other more
> populous, more intelligent races -- we're still
> toast). Thus far, on the AGI mailing list, no one
> has been able to propose a scenario where humanity
> (or an entity) is destroyed or abused where it is
> not either *clearly* contradictory to the
> formulation or *clearly* ethically permissible for
> it to be destroyed/abused. This list is now an acid
> test to see if anyone here can come up with such a
> scenario.
> Since I believe that crossing ethical boundaries for
> the sake of our safety is a *BAD* idea (because
> making this allowance for ourselves actually
> diminishes our safety because it effectively gives
> others permission to cross this boundary as well --
> i.e. not be Friendly), I strongly suspect that it
> *is* formally provable that we cannot be safer than
> "as safe as ethically possible".
> It seems obvious in hindsight but it was an epiphany
> when I realized while working on this that it would
> be highly beneficial to *me* to officially declare
> myself to be a Friendly entity (according to the
> formulation) since doing so was only in my own
> self-interest. I have done so on the AGI list and
> am doing so again below.
> The coolest thing is that I think that the core of
> these ideas can be captured in a single catchy
> slogan "Friendliness: The Ice-9 of Ethics and the
> Ultimate in Self-Interest". Given the probable
> existence of super-Friendliness, it's a bit of
> hyperbole but I think that it's fully justified to
> promote the spread of Friendliness (which is in
> *all* of our best interest).
> There have been some fun discussions going on
> through the AGI list
> (
> including a great alien contact scenario thread
> titled 'A Few Questions for "Vladimir, the
> Destroyer"' and I am inviting any and all to take a
> swing at my strawman (just make it a competent swing
> :-).
> = = = = = = = = = =
> The Friendliness Declaration
> Hi, I’m Mark Waser.
> I am hereby declaring myself to be a Friendly entity
> as of March 12, 2008. I have not
> knowingly/intentionally performed any UnFriendly
> actions since this date.
> I have not previously been declared UnFriendly by
> any declared Friendly entity. I have not been
> requested to put any disputed actions or
> unintentional UnFriendly actions in this declaration
> by any declared Friendly entity.
> I am voluntarily undertaking the goals of
> Friendliness because I understand that doing so is
> in my own best self-interest and is the best way to
> achieve all my other goals.
> I am voluntarily undertaking the primary/overriding
> goal of Friendliness as my primary and pre-eminent
> goal. I understand this goal to be:
> Do not act contrary to someone’s/anyone’s goals
> unless absolutely necessary for the fulfillment of a
> reasonable/rational personal goal (explicitly not
> including generic sub-goals like money, power,
> pleasure, religion, etc.)
> I am voluntarily undertaking the goal of spreading
> Friendliness and assigning it a reasonable priority
> (i.e. enough to be a net asset to a Friendly
> Society) because I understand that doing so is in my
> own self-interest.
> I will act in accordance with these goals according
> to the rules/laws of a Friendly Society included in
> this declaration. I understand that if I knowingly
> and intentionally act contrary to these rules/laws
> that I can be declared UnFriendly and will need to
> make another declaration of Friendliness including
> information about the most recent times that I have
> been declared UnFriendly.
> I understand that the rules/laws of a Friendly
> Society include the requirement that all statements
> in this declaration be truthful under penalty of
> having this declaration declared null and void and
> my being declared UnFriendly for attempting to
> impersonate a Friendly entity.
> In particular, I understand that hiding or lying
> about any UnFriendly actions (even if
> unintentionally or unknowingly performed) is
> contrary to the rules/laws of a Friendly Society. I
> understand that while I may defend myself against
> unfair charges of UnFriendly actions, I must admit
> to all UnFriendly actions (even if unknowingly
> and/or intentionally performed) and must include all
> disputed charges in this declaration as requested
> for a reasonable length of time or risk having this
> declaration declared null and void and my being
> declared UnFriendly.
> I understand that the rules/laws of Friendly Society
> state that declaring another entity UnFriendly is,
> in itself, UnFriendly unless there is reasonable
> cause for doing so, I have executed a reasonable due
> process, and am sure beyond reasonable doubt. I
> understand that the rules/laws further require that
> I retract any declarations of UnFriendliness if
> *any* reasonable doubt subsequently arises and I
> must volunteer my most recent retractions of
> UnFriendliness declarations (if any) when asked and
> in any subsequent UnFriendliness declarations.
> <signed>
> Mark R. Waser
=== message truncated ===

Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:02 MDT