42-No-Entitlement.txt

TITL:
   *No Entitlement or Choice Respect*
   By Forrest Landry,
   November 18th, 2022.

ABST:
   Examines and completely rejects
   an argument that attempts to
   justify AGI Arms Race dynamics.

PREF:

   A lot of the following content
   was evoked and influenced by and/or
   is directly quoting of (@ this https://twitter.com/KerryLVaughan/status/1536364299089854471)
   Kerry Vaughan Tweet thread.

   Any responsibility for selecting, editing,
   and extending well beyond the intended meaning
   of his texts is my own, and has been done
   *without* his permission, knowledge,
   or advance notification.

   Disclaimer: I have no personal knowledge of Kerry Vaughan
   or of his interests/involvements, intentions, etc.
   I represent only myself and have no known social
   graph overlap or vested interests in common with his.

INTR:

   My commentary and extensions are that of someone
   who is strictly an outsider, representing his own
   research into AGI alignment/safety impossibility.
   I came across this tweet and saw it as a way to
   use it as a base to extend it into a more formal
   and complete expression of what was mutually worth
   objecting to, based on my own independent arguments
   and observations.

TEXT:

   My overall assessment of the situation with AGI
   safety research largely agrees with the following.

   > It is (or should be) very obvious that AGI labs
   > (eg OpenAI, DeepMind, and others)
   > are THE CAUSE of the fundamental problem
   > the AI Safety field faces.

   > To vastly oversimplify the situation,
   > you can think of AI Safety as a race:.
   >   - In one corner you have the AGI builders
   >   who are trying to create AGI as fast as possible.
   >   - In the other corner,
   >   you have people trying to make sure
   >   that AGI will be "aligned" with human goals
   >   once we build it.

   It is to be noted that "alignment" is often
   equivocated to be "with the interests of the makers
   and the owners of such systems" rather than
   "for the good of humanity, etc".
   This is, however, another issue and will not be
   considered further here.

   The "reasoning" usually being proposed
   for even allowing a 'race' in this form
   to occur anwhere in the world at all,
   is some variation of the following:.

     > (1) If AGI gets built
     > (2) before we know how to align it,
     > (3) it *might* be CATASTROPHIC; however
     > (4) it *might* be the case
     > that aligning an AGI *might* be *maybe* possible,
     > and that, (5) if given enough time and effort,
     > we *might/maybe* eventually solve
     > the alignment/safety problem.
     > (6) So we should (we recommend that we) build AGI first,
     > so that (7) *maybe* we will have enough time
     > to solve that alignment problem,
     > and then *maybe* (8) we can use "our" AGI
     > to prevent other "bad aligned AGI"
     > from being made by those other
     > (9) maybe evil companies".

   Unfortunately, nearly everything in this pseudo argument
   is simply wrong.  And overall, the situation of the game
   is even worse than gambling, and should be made even more
   illegal than the use of playing "roulette" with nukes.

   Re 1; Will AGI get built?  Very likely yes, given the
   investor and corporate social proclivity to do so,
   belief in near unlimited profits, strong moat, etc.
   Actual social back-pressure applied at this point maybe
   the only actual place that is actually actionable.

   Ie, if we are going to stop near certain death
   to the planet very much sooner than its natural time,
   we will need to put together strong social sanctions
   to stop AGI development or around this first step.

   Re 2; what are the chances that AGI will be made
   before a method of alignment is discovered?
   It is 100% certain --
   there is actually no possible *lasting* method of AGI alignment.
   It may be possible to setup some method that *maybe*,
   for a little while, looks like AGI alignment
   for some short term, but even this is most likely
   just an illusion of alignment and safety,
   and not actual alignment or safety.
   That true lasting AGI safety is strictly impossible
   on just the basis of the math and physics alone.

   Re 3; will AGI be catastrophic?
   Yes, eventually, for sure, it will.
   There is no "maybe" here.
   For all reasonable definitions of "catastrophic"
   that are not marketing spin of attempting to covertly
   redefine the word "catastrophic"
   to mean something fake like
     "maybe bad for future technology"
   or
     "bad for future potential" corporate private profits.

   They may sometimes say or suggest "human potential",
   but the notion of 'human' is spun into 'machine',
   and building machines makes someone money,
   so it is actually, functionally the same as,
   in all cases 'bad for future potential profits'.
   That is the real meaning, hidden behind clever
   rhetoric and social media marketing, broadcasting,
   book writing, glossy print websites, blogs,
   podcasts, etc, etc.

   'Catastrophic' will remain the unequivocal loss
   of all or near all natural organic life,
   inclusive of all humans,
   whole world ecosystem destruction, etc,
   or some significant fraction thereof,
   as known and assessed per/by any true random selected
   actual diverse consensus, etc.

   Re 4; this is just plain wrong.
   It is simply never going to be the case
   that aligning an AGI "might" be "maybe possible",
   whether on the basis of principle or on *any* practical basis,
   at all, ever.
   The fact of "alignment" and "safety"
   to natural organic life/humans
   is fully and actually structurally impossible, by definition,
   due to the simple fact of the AGI being both _artificial_
   and agentic (ie, self authoring, etc).

   Re 5; this is also just plain wrong:
   no investment of time and effort, no matter how much,
   how extensive, etc, will ever accomplish
   the actually logically and physically impossible.
   To invest in the delusion of it being otherwise
   is simply to continue to be in waste.

   The only reason that anyone would knowingly
   engage in such waste
   (and the vast number of people are simply unknowing,
   or maybe even just indifferent, numb, traumatized, etc)
   is if they were executing or involved in, directly or
   indirectly, some sort of con, scam, or Ponzi scheme --
   ie, nearly any corporation, nar-social person, etc.

   That no amount of time/effort/creativity "invested"
   in maybe aligning AGI will do anything other than
   social signal marketing, spin, and hype.

   Re 6; as the conclusion that they want you to believe,
   however they can get you to, is the center of the scam,
   the essence of the con "you too can get rich quick" and
   "it will be good for you too", etc,
   along with all sorts of promises of "you can live forever,
   and be forever happy, as an upload into this
   perfect AGI supported virtual reality",
   and other utopian dreams
   used to drive people into a cult.

   This makes EA and the alignment research groups
   on a strict functional and practical operations basis,
   more like a cult that cannot not seem to be pretending,
   really really hard, with all of the reason and reasons
   and lots of logical and rational sounding arguments,
   that they are not a cult.
   They will for sure protest,
   and for sure it will be far too much.
   The traumatized and Stockholm Syndrome people are often
   the least trustworthy defenders of their own sovereignty,
   dignity, and safety.

   Engineers and researchers in an environment saturated
   with hidden marketing and spin, bad social media process,
   and consistently failing to be recognized as such,
   when the few unfeeling nar-social executive elites
   continue to drive the hype towards "wonderfulness"
   and "something for everyone" -- it is all symptomatic.

   Building an AGI will make someone money,
   and for sure make someone forever (in) famous.
   And it sounds/seems really cool too!

   Of course the nar-social executive elite and VC project
   doners, investors, grant writers, etc,
   will recommend something that is to their extensive
   seemingly unlimited profit benefit,
   hidden by perfected power and control,
   via AGI machines with covert programming mixed in with
   all of the other inscrutable and non-inspectable
   neural net "logic".
   The return on their investments,
   in their own lifetime,
   and with maybe also no actual care or concern
   about anyone else,
   now or ever in the future,
   will be so extensive
   that they conveniently overlook
   the absence of a capability
   and a right to choose,
   on everyone else's part.
   The net effect is that everyone else's
   right to choose,
   will be permanently disrespected (ie; destroyed).

   "Being first" is a distraction
   and uses gaming marketing to re-direct your attention
   and to entrap you in their context and way of thinking.

   Re 7; "we" will never have "enough" 'time'
   (or resources, or capability, or...etc),
   to align AGI since it is *still* actually impossible to do so.
   Of course, while *you* are finding out the hard way,
   *they* as the scam artist, will be making mega-bank!
   Of course, you will never know that (never will see $).
   All you will notice is that things get increasingly
   hard and harder for you and everyone you know,
   everywhere else in the world, except for that one
   seemingly super perfect entrepreneur super being
   that is doing very very well: "killing it" of course!.

   Re 8; and _even_if_ the above were not enough,
   it has also been shown that _even_if_
   we were to assume, _against_reason_,
   but for the sake of principled debate,
   that we assume that "we" "had an aligned AGI"
      (of course, nearly all safety arguments start with
      that premise of assuming what they want to prove,
      which is like assuming that divide by zero
      is a "valid step" (NOT!) in 'any proof' --
      ignoring completely that by assuming the false to be true,
      even once, we can therefore get to any conclusion "truth"
      that we want to -- but ignoring that insanity)
   that even this assumption is not at all enough
   to also ensure that no one else develops and deploys an AGI
   (regardless of whether that AGI is aligned/safe or not).
   Ie, that "having" one AGI does not exclusively prevent
   anyone else from also "having" an AGI,
   short of *also* assigning and assuming
   near God like powers and capabilities
   to "our" favored AGI.

   Basically, anything that is strong enough to protect us
   is also strong enough to kill us, and in this case,
   to also strong enough to for sure _not_care_ at all
   about protecting us.

   Where given the near total artificialness of AGI,
   that it will also have a complete absence of care,
   then the net result is the very essence of
   a complete absence of alignment and safety.
   That is just what those words actually mean.
   The net effect is that "idea 8" also completely fails.

   Re 9; By speculating that all other
   companies/people/groups of the world
   are also 'maybe evil' is simply a revealing of
   a typical nar-social bias.
   Given that "we"...
     (narcissists generally do not distinguish between
     self and other --
     that all "others" are actually appendages of themselves,
     a bit like other unfeeling and unalive 'accessories' etc)
   ...are 'secretly evil'...
     (internally, nar-social people notice that they
     privately do not actually care
     about respecting/furthering anyone else's choices,
     though they do also for sure notice
     that it is convenient and helpful to social signal
     _as_if_ they do care, etc,
     as long as such signaling actions
     are not too costly/effortful)
   ...then "we" will expect that all other actors/agents
   in the world
   are also actually 'evil'
     (where the notion of 'evil' is 'that which does not further',
     which in essence, is does not further
     choice, love, life, etc, for anyone else,
     anything in the natural ecology, ecosystem, etc).


   > while OpenAI and DeepMind may publically (currently)
   > appear to be advancing the cutting edge the most,
   > this critique applies to *any* company or academic researcher
   > that spends their time working to solve
   > the bottlenecks to building AGI.

   > AGI labs like OAI and DeepMind have it as their MISSION
   > to decrease the time we have.

   > Their FOUNDING OBJECTIVE is to build AGI
   > and they are very clearly and obviously trying
   > *as hard as they can*
   > to do just that.
   > They raise money, hire talent, etc,
   > all premised on just this goal.

   Of course, such single minded dedication is obsession,
   and it is an obsession that is blind to all else
   that is of value in life.  The pursuit of profit,
   maximized and hidden behind lots of looks-good/feels-good
   (but ultimately actually wrong/evil) logic/argument remains.

   > Every day an AGI engineer at OpenAI or DeepMind
   > shows up to work and tries to solve the current bottlenecks
   > in creating AGI...
     ...we make the long future less likely
     to be anything other than
     the total and permanent extinction
     of all human beings, all life on this planet.

     Of course, all of those engineers and researchers
     have been sold the 'cool-aid' of working on
     "the most interesting problems" and "the best chance"
     of "helping humanity".  All corporate marketing
     and cult insider speak to get the rank and file
     to do the bidding of the investment masters.
     No one ever fights a war against hard odds to kill
     unless they have been convinced by someone first
     that such action is necessary and inevitable,
     and in that sense, somehow "good".

   > My very bold take is that THIS IS BAD!

   Of course, the trans humanists, Bostrom, etc,
   will attempt to tell you (argue with you, and they are
   oh so much smarter than everyone, so surely they have
   actually thought this out (to whose benefit?))
   that all of the above is somehow "good" and
   "necessary for the future" and that we should simply
   "submit" to the "inevitable development of life harmful tech".
   Dominance and submission is the essence of the tech world.
   They and others suggest that it would be a good idea
   for _both_ the natural world and the organic human world
   to get replaced by an artificial machine world.
   Forever in the future.

   Of course, this is so maximally disrespectful
   of all of the real and embodied choices
   of literally every other creature on the entire planet.
   The level of hubris and arrogance is staggering.
   That only their choices matter, and they will take and
   steal your own choices, your own dignity, sovereignty,
   and safety, not only the choice as to whether or not to AGI,
   but also of whether or not to have a future
   with organic life in it at all.
   That all future choices will be "owned" by
   the these new forms of artificial hardware
   of the AGI substrate systems.

   OF course, it will be noticed that "we"
   can then also skip even
   'simulating of virtual "happy" "humans"'
   with some sort of number counter,
   since that is also just a simulation of a virtual human,
   though one at a massively lower level of fidelity.

   Do we need to simulate both the full neural net
   of a maybe real look alike person
   and also the total world environment that they live in?
   That is a lot of effort! No, lets instead simulate
   just the experiences of that environment
   inside their own virtual minds.
   Of course, may as well skip even that
   and simply wire-head them --
   set the happiness bit to always be on.

   Of course, eventually, the machines will notice
   that even that too is just wasted space,
   wasted computation, energy, etc, and will move on
   to lower fidelity of just simply considering
   how many simulated happiness bits are set.
   So now all of the future of "humanity"
   is replaced by a happy number counter.
   Except that even virtualized virtual "humans"
   represented by a number simply has no merit
   in the post modern machine economy/ecology,
   so we can skip that too.

   So the perfected trans-humanist outcome
   is simply no "human" or 'humans' at all.
   Given that there is actually zero respect
   of any human choice, life, or embodied being
   by successive simplifications of
   what it even means to be human,
   to the point of there being not even
   a simulation of fake choices,
   that in all cases both the means and the ends
   are a total absence of choice, freedom, and value.

   It is an absolute total loss.
   There is no more pretending that there is a moral
   justification, via a faulty total utilitarian logic
   of the loss of *real* choices of *real* people today
   against the probability of 10E+50 fake "humans"
   is actually zero, because they are not human,
   do not make real choices, have zero probability
   of even occurring in any of the proposed futures
   that are anything other than dreaming, marketing,
   spin, etc, etc.

   Any number multiplied by zero is zero,
   so the "technological long termism"
   is simply a bid for death.

   > But Demis Hassabis and Sam Altman
   > are not psychopaths or morons.

   Whether or not those two people --
   I do not know them at all --
   are actually internally privately uncaring about others,
   the actual functional effect of their actions embodies
   and encodes the fact of 'not caring about or respecting
   the choices of others'.  It is not about what they are,
   it is about what they do.  I do not need to know them
   to see the real effects of their choice of actions,
   value, and life priority.

   Their choices are not respecting other people's choices.
   That is already the case, now, today.
   No other person/group in the entire world, now
   or forever into the future,
   is being considered or respected in
   the choice of whether or not Hassabis/Altman
   are directing effort/resources to/towards making AGI.

   > If they get close to AGI without solving alignment
   > they can just not deploy the AGI.

   Moreover, it is unclear, regardless of their possible
   level of intelligence, whether they (Sam and Demis)
   will even have the choice as to whether or not
   to use/deploy that AGI, once it is made.
   Social corporate politics, along with
   market forces driven by investors
   and the current corporate favoring legal structures
   of many countries (USA primarily)
   may drive the company social group.

   It is especially ironic that these same people,
   (or the engineers who also drank the 'Kool aid')
   will argue on one hand that solving large scale
   social coordination problems is impossibly hard,
   and yet will then, nearly immediately and at their
   own lopsided logic suggest that a single person
   can simply "decide" to "not deploy" what some
   large investment group with significant power
   (but a lot less understanding of the issues involved)
   might simply 'tell them to do it', regardless of
   their "concerns", since after all, we invested
   all of this money, not for nothing!.
   So who is responsible, the person who issued the
   direct order, or the one who foolishly followed it?
   Just like the Nazi army all over again.
   Trans-humanism and total utilitarianism is
   the new eugenics, and for all of the same reasons.

   Moreover, even just one defecting engineer
   within that company, working secretly at night
   can be tempted or coerced by some outside company
   or force to "turn it on and just try it",
   regardless of the risks.  Maybe some janitor
   just wants to impress their maybe girlfriend.
   She will have no understanding that by his action,
   she will never have the option to have her maybe children
   live in a safe world as a result.  Does anyone really
   expect that the woman present will be able to "tell"
   some testosterone driven man to "please stop short of
   courting danger" to impress her with his intelligence?

   Maybe the AGI itself then decides for all of us,
   also not caring about any of our human choices
   (being inherently artificial and all)
   that its not-natural outcomes will be loss of nature
   (ie, death).  It was only supposed to be partially
   tested, in a "contained" environment, and yet that
   was not even close to being actually an option
   for us to prevent disaster.

   In any case, the probability of 'if it is made at all,
   it will very likely be used' is a very strong factor --
   especially in an experimental culture
   (as corporations of this type tend to set up)
   and with a total absence of any regulation
   or legal or social system implying any level of culpability
   or even guilt at all (scientists and researchers are
   supposed to be above all of that emotional or ethical
   integrity -- only the truth matters, let someone else
   figure out the ethics of the gain of function research).
   Who is to blame when a machine kills a person?
   Blame the gun maker or the gun builder?
   Our legal system does not even have one single law against
   having an 'Automatic-Assisted Genocide Initiative' (AGI),
   there is no case history, no consensus agreed upon
   common law assumptions, etc.


   > Every. Single. Day. the AGI labs
   > are steadily advancing
   > the state of the art
   > on building AGI.

   > With every new study they publish,
   > with every researcher they train,
   > and any AI technology that they commercialize,
   > they also make it easier for every other AGI lab
   > to build and deploy an AGI.

   > So unless they can somehow refrain...

     (in the same sense as attempting to refrain
     from committing a crime is so very difficult,
     ie; as assuming falsely that *everyone*
     will for sure commit a crime
     when given so much temptation, etc,
     as if the nar-social assumes that everyone
     is nar-social, immoral, etc, like them, etc).

   > ...from deploying an unaligned AGI...

     Especially when *all* AGI will for sure be,
     eventually unaligned, unsafe, catastrophic.
     (But of course, it is so deeply unfashionable
     to say that when your research paycheck
     is being provided to you by the same cult).

   > ...and stop EVERYONE ELSE from doing the same,
   > they continue to be in the business
   > of robbing humanity of valuable time.

     Of robbing everyone of a future,
     of stealing the value of all future choice
     so that they might be a little more fashionable
     in their mega-yacht in the maybe near term.

     Of course, the question of who is to stop
     the crimes of others if/when you are yourself
     actually doing that same crime, is not so easy.

   > They, and anyone else working on AGI development,
   > regardless of which company they are in
   > or which national legal jurisdiction they are in,
   > are the cause of the fundamental problem
   > faced by the AI Safety community.

   Ie, the general confusion of "can we make AGI safe"
   and the continued failed attempts to do so
   are being mistaken as (are being spun/marketed as)
   "not trying hard enough", (make you feel guilty
   and put safety researchers on the defensive, etc)
   as a means to divert attention from
   the (mis-) perceived short term high return
   profit making activities
   that they can do in the interim.

   > In conclusion: Stop building AGI you fucks.

   Amen.


   > Notably, a number of people in the AI Safety community
   > basically agree (in private) with all of this
   > but think that no one should say any of this.

   The choice of so many AGI safety engineers/researchers
   to be tacitly silent is the outcome of
   a real lack of respect for their own choices, and freedom,
   to act and speak in accordance with their own values,
   that they have internalized as a kind of trauma
   due to the exposure to an ambient nar-social environment.
     (a social environment created by things like social media,
     total ambient marketing, the whole world being eaten by
     market process, crypto scams and cons, scams, extortion,
     no corporations being held actually accountable to their
     communities, pollution, environmental harms
     to the land and seas, etc).

   Someone once mentioned to me that hardly anyone on the
   "effective altruism" forum and/or the "alignment forum"
   is actually a friend of anyone else on EA, AF, etc.
   A more significant indictment of a lack of health and vitality
   and connection in that "community" could not asked for.
   All of the engineers and researchers are atomized,
   constantly fighting amongst themselves
   with overwrought ideas and concepts, and arguments
   which are all the tools of the nar-socal elite
   hiding in plain sight among them,
   looking to make a quick buck (more crypto anyone?)
   via some sort of technological miracle --
   a perpetual benefit machine to your glory,
   my king.

   In a scam/con, it does not matter if the tech miracle is real,
   or is safe, useful, culturally beneficial etc.  It only matters
   that at least some few people believe in it, and can be used
   to promote the ideas and arguments supporting the hype.
   Out of this is what Ponzi schemas are made of.

   The FTX situation was a wake-up call.
   And of course, it will eventually be ignored --
   the "lessons internalized" will be the wrong ones,
   and the hype cycle of the next "lets build an AGI"
   will continue.  Cults and cult leaders never learn.
   Although they can be very street smart,
   there is no assurance that they will learn
   love of life and emotional integrity.
   The only thing we can hope for is
   that the rest of us *do* learn to recognize
   and identify the action of greed, the nar-social,
   and the effects of their action, and thus,
   become wiser and more savvy to their manipulations,
   setting hard boundaries against what is not ok.

   We need to forever read and understand the acronym
   "AGI" as standing for "Authorized Genocide Initiative",
   one that is against all life, all freedom,
   all that is human and organic and wild and beautiful,
   to replace it with dead profit in the form of numbers
   in some bank account database.
   Bits set there are just waiting for the power to go out
   and become truly meaningless
   in the atomic sunset of humanity.