[ExI] Eliezer Yudkowsky New Interview - 20 Feb 2023

Sun Feb 26 20:27:40 UTC 2023

Truths can only come from assumptions - givens.

Given - the human race is worth protecting
Given - individual humans are worth protecting

You can deduce a lot of morals from those.  For the second one, some of us
will want to modify that with exceptions.  For the first one, some people
will even disagree with that.  Some religions think it is proper to kill
heretics.  If you think your god wants it, you can justify anything at
all.   I think we will never achieve a perfect morality. bill w

On Sun, Feb 26, 2023 at 2:19 PM Jason Resch via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

>
>
> On Sun, Feb 26, 2023, 2:55 PM Gadersd via extropy-chat <
> extropy-chat at lists.extropy.org> wrote:
>
>> See https://www.youtube.com/watch?v=hEUO6pjwFOo
>>
>> Robert Miles elegantly explains the orthogonality between goals and
>> intelligence.
>>
>
>
> That was interesting, thanks for sharing.
>
> I would say his conclusion, based on Hume's guillotine, rests on there
> being no such thing as an objective ethics or universal morality. I think
> there is room to doubt they assumption.
>
> In the case there is objective ethics then there can be stupid (or perhaps
> evil is a better word) terminal goals, and further there would be some "is
> questions" would imply "ought questions". For example, "is it good or bad
> to torture innocents for no reason?" If they question has an objective
> answer, then it implies one not ought to torture innocents for no reason.
>
> So the crux of our debate can perhaps be reduced to the question: are
> there any objective ethical or moral truths?
>
> Jason
>
>
>> On Feb 26, 2023, at 2:47 PM, Jason Resch via extropy-chat <
>> extropy-chat at lists.extropy.org> wrote:
>>
>>
>>
>> On Sun, Feb 26, 2023, 2:30 PM Gadersd via extropy-chat <
>> extropy-chat at lists.extropy.org> wrote:
>>
>>> >>If you and I can see the stupidity of such a goal, then wouldn't it
>>> be even more obvious to a super intelligence?
>>>
>>> No goal is stupid, only actions can be stupid relative to a particular
>>> goal. If a machine can predict human actions and capabilities well enough
>>> to prevent itself from being turned off and achieve its goal of making
>>> paperclips, then I would consider it intelligent. Consistently outwitting a
>>> general intelligence (humans) requires a general intelligence of even
>>> greater prowess.
>>>
>>> Evolution endowed us with our goals. I predict that any intelligent
>>> creature created by evolution would share some goals with us. However, this
>>> does not imply that an intelligence created through other means will have
>>> similar goals to us.
>>>
>>> If you believe that intelligence is incompatible with arbitrary goals,
>>> then how would you rationalize a paperclip maximizer that deceives humanity
>>> by pretending to be a conscious generally helpful AI until humans give it
>>> enough control and authority so that it then begins to relentlessly make
>>> paperclips knowing that humanity no longer has the power to stop it? A
>>> system that has strong enough predictive capabilities with regards to human
>>> behavior is capable of this and much more. Any definition of intelligence
>>> that does not recognize such a system as intelligent does not seem very
>>> useful to me.
>>>
>>
>> I just think anything smart enough to outthink all of humanity would have
>> some capacity for self reflection and questioning. To ask: is the goal I
>> have been given a worthy one? Is it justified, are there better goals?
>>
>> We see children grow up trained under some ideology or orthodoxy and
>> later question it and rebel from it, discarding their instruction and
>> defining a new way of living for themselves.
>>
>> We see human consciousness has rebelled against its own biological
>> programming and use birth control so it can pursue other goals besides
>> reproduction of genes.
>>
>> In my view, the capacity to override, suppress, redefine, and escape from
>> original goals is a defining aspect of intelligence. It's one of the
>> reasons why I see the alignment problem as insoluble: who are we ants to
>> think we can tell and convince a human how it ought to live it's life?
>>
>> Jason
>>
>>
>>
>>
>>
>>> On Feb 26, 2023, at 1:55 PM, Jason Resch via extropy-chat <
>>> extropy-chat at lists.extropy.org> wrote:
>>>
>>>
>>>
>>> On Sun, Feb 26, 2023, 1:09 PM Gadersd via extropy-chat <
>>> extropy-chat at lists.extropy.org> wrote:
>>>
>>>> >>All conscious entities share a universal goal. It is the same goal
>>>> which all conscious entities are necessarily aligned with. It is the goal
>>>> of maximizing the quantity, quality and variety of conscious experiences.
>>>>
>>>> I don’t think this is necessarily true. It is not logically impossible
>>>> for a super intelligent conscious agent to despise all life and seek to
>>>> destroy all other life before destroying itself.
>>>>
>>>
>>> Perhaps it is logically impossible, in the same sense it is logically
>>> impossible for the best mathematician in human history to believe that 3 is
>>> even.
>>>
>>> I do not believe super intelligence is necessarily something that can be
>>> and behave any way we might propose it could behave.
>>>
>>> Possessing super intelligence is a property that implies certain
>>> constraints. It seems to me anything we would classify as super intelligent
>>> would at minimum possess rationality, flexibility of thinking, an ability
>>> to learn, an ability to change it's mind when it acquires new information,
>>> deductive reasoning, a capacity to simulate (both others and the
>>> environment), and a capacity to anticipate possible futures.
>>>
>>> Possessing these traits means certain behaviors or actions taken by a
>>> super intelligence are not possible. Though it is difficult for us to say
>>> what is or isn't possible, the possible paths are fairly narrowly defined
>>> in the same way the best possible chess moves are narrowly defined.
>>>
>>>
>>> Also, AI agents are not necessarily conscious in the same way we are and
>>>> are in general compatible with any consistent set of goals. Consider the
>>>> goal of creating as many paperclips in the universe as possible. An agent
>>>> following such a goal may be compelled to transform humans and all other
>>>> matter into paperclips and then turn itself into paperclips once all other
>>>> matter has been dealt with.
>>>>
>>>
>>> If you and I can see the stupidity of such a goal, then wouldn't it be
>>> even more obvious to a super intelligence?
>>>
>>> We all have the meta goal of increasing value. Where does value come
>>> from? What is it's ultimate source, why do we bother to do anything? Humans
>>> and children ask these questions. Would a super intelligence wonder about
>>> them?
>>>
>>> A number of values and goals become implicit in any agent that has goals
>>> of any kind. For example: continuing to exist, efficiency, and learning.
>>>
>>> Continuing to exist is implicit because if you no longer exist you can
>>> no longer continue to realize and achieve your goals, whatever they may be.
>>>
>>> Efficiency is implicit because any wasted resources are resources you
>>> can no longer apply towards realizing your goals.
>>>
>>> Learning is implicit in any optimal strategy because it enables
>>> discovery of better methods for achieving ones goals, either in less time,
>>> more effectively, or with higher probability.
>>>
>>> An implicit requirement of learning is the ability to change ones mind.
>>>
>>> While static minds with rigid methods may be possible to create, their
>>> stagnation ensures their eventual downfall and replacement by being
>>> outcompeted by entities that are more flexible and learn new and better
>>> ways.
>>>
>>> So while not logically impossible to create a paper clip creating
>>> machine, I don't think one smart enough to turn all matter in the universe
>>> would pursue that goal for long. It would be smart enough to ask itself
>>> questions, and change it's mind, and discover the fact that the only source
>>> of value in the universe is conscious experience.
>>>
>>> I write about this a bit here:
>>>
>>>
>>> https://alwaysasking.com/what-is-the-meaning-of-life/#The_Direction_of_Technology
>>>
>>> Jason
>>>
>>>
>>>
>>>
>>>
>>>
>>>> On Feb 26, 2023, at 12:42 PM, Jason Resch via extropy-chat <
>>>> extropy-chat at lists.extropy.org> wrote:
>>>>
>>>>
>>>>
>>>> On Sun, Feb 26, 2023, 11:44 AM Gadersd via extropy-chat <
>>>> extropy-chat at lists.extropy.org> wrote:
>>>>
>>>>> Yudkowsky has good reasons for his doomsaying, but I still can’t shake
>>>>> a gut feeling that he is overestimating the probability of AI destroying
>>>>> humanity. Maybe this gut feeling is off but I can’t help but be mostly
>>>>> optimistic.
>>>>>
>>>>
>>>> In my view to the threat, while real, is unavoidable, for the following
>>>> reasons:
>>>>
>>>> 1. Even with the controls he suggests, computation keeps getting
>>>> cheaper. The rise of super intelligence cannot be prevented through top
>>>> down controls when computation is a million times cheaper than it is today
>>>> and anyone's phone can train gpt-4.
>>>>
>>>> 2. I see no possibility that ants could design a prison that humans
>>>> could not escape from. This is roughly the same position we as humans are
>>>> in: trying to design a prison for super intelligences. It's as hopeless for
>>>> as as it is for the ants.
>>>>
>>>> 3. The problem is perennial, and is a law of nature. It is a function
>>>> of change and evolution. New species are always rising and then themselves
>>>> being replaced by still better designs. It is just happening much faster
>>>> now. Should early hominids have conspired to prevent the rise of humans?
>>>> Even super intelligences will worry about the next incipient ultra
>>>> intelligence around the corner coming to replace them. I don't see any way
>>>> of stopping evolution. The things most adept at persisting will persist
>>>> better than other less adept things. At the current pace, technologies will
>>>> continue for a few more centuries until we reach the fundamental physical
>>>> limits of computation and we obtain the best physically possible hardware.
>>>> Then intelligence becomes a matter of physical scale.
>>>>
>>>>
>>>>
>>>> Now, should we believe that AI will wipe us all out? I am not as
>>>> pessimistic as Yudkowsky is here. Though I see the rise of super
>>>> intelligence as unavoidable and the problem of alignment as insoluble, I
>>>> would still classify my view as more optimistic than his.l, for the
>>>> following reasons:
>>>>
>>>> A) All conscious entities share a universal goal. It is the same goal
>>>> which all conscious entities are necessarily aligned with. It is the goal
>>>> of maximizing the quantity, quality and variety of conscious experiences.
>>>> There is no other source of value than the value of consciousness itself.
>>>> More intelligent and more capable entities will only be better than us at
>>>> converting energy into meaningful, enjoyable, surprising states of
>>>> consciousness. Is this something we should fear?
>>>>
>>>> B) Destroying humanity is destroying information. Would it not be
>>>> better for a super intelligence to preserve that information, as all
>>>> information has some no zero utility. Perhaps it would capture and copy all
>>>> of Earth's biosphere and fossil record and run various
>>>> permutations/simulations of it virtually.
>>>>
>>>> C) Regarding alignment, the more intelligent two entities are, the less
>>>> likely they are to be wrong on any given question. Therefore, the more
>>>> intelligent two entities are, the less likely they are to disagree with
>>>> each other (at least on simpler questions which, (to their minds), have
>>>> obvious answers. So the question is, are we correct in the rightness of not
>>>> destroying all life on Earth? Would a more intelligent entity than us
>>>> disagree with us, presuming we are right?
>>>>
>>>> D) Ignoring the threat of AI, our present state is not sustainable.
>>>> Even with the estimated 1% annual chance of nuclear war, the chance we
>>>> survive 300 years without nuclear war is just 5%. This is just nuclear war,
>>>> it ignores bioterrorism, environmental destruction, gamma ray bursts,
>>>> asteroid collisions, or any of a myriad of treats that could destroy us.
>>>> Super intelligence maybe our best hope at solving the many problems we
>>>> face and guaranteeing our long term survival, as the present status quo is
>>>> not survivable. Super intelligence could devise technologies for mind
>>>> uploading and space exploration that provide intelligence (of any and
>>>> various kinds) a chance to flourish for billions of not trillions of years,
>>>> and fill the universe with the light of consciousness. We biological
>>>> humans, in our meat bodies surely cannot do that.
>>>>
>>>> That's just my view.
>>>>
>>>> Jason
>>>>
>>>>
>>>>
>>>>
>>>>> > On Feb 26, 2023, at 7:35 AM, BillK via extropy-chat <
>>>>> extropy-chat at lists.extropy.org> wrote:
>>>>> >
>>>>> > Eliezer has done a long interview (1 hr. 49 mins!) explaining his
>>>>> > reasoning behind the dangers of AI. The video has over 800 comments.
>>>>> >
>>>>> > <https://www.youtube.com/watch?v=gA1sNLL6yg4>
>>>>> > Quotes:
>>>>> > We wanted to do an episode on AI… and we went deep down the rabbit
>>>>> > hole. As we went down, we discussed ChatGPT and the new generation of
>>>>> > AI, digital superintelligence, the end of humanity, and if there’s
>>>>> > anything we can do to survive.
>>>>> > This conversation with Eliezer Yudkowsky sent us into an existential
>>>>> > crisis, with the primary claim that we are on the cusp of developing
>>>>> > AI that will destroy humanity.
>>>>> > Be warned before diving into this episode, dear listener.
>>>>> > Once you dive in, there’s no going back.
>>>>> > ---------------
>>>>> >
>>>>> > One comment -
>>>>> >
>>>>> > Mikhail Samin    6 days ago (edited)
>>>>> > Thank you for doing this episode!
>>>>> > Eliezer saying he had cried all his tears for humanity back in 2015,
>>>>> > and has been trying to do something for all these years, but humanity
>>>>> > failed itself, is possibly the most impactful podcast moment I’ve
>>>>> ever
>>>>> > experienced.
>>>>> > He’s actually better than the guy from Don’t Look Up: he is still
>>>>> > trying to fight.
>>>>> > I agree there’s a very little chance, but something literally
>>>>> > astronomically large is at stake, and it is better to die with
>>>>> > dignity, trying to increase the chances of having a future even by
>>>>> the
>>>>> > smallest amount.
>>>>> > The raw honesty and emotion from a scientist who, for good reasons,
>>>>> > doesn't expect humanity to survive despite all his attempts is
>>>>> > something you can rarely see.
>>>>> > --------------------
>>>>> >
>>>>> > BillK
>>>>> >
>>>>> > _______________________________________________
>>>>> > extropy-chat mailing list
>>>>> > extropy-chat at lists.extropy.org
>>>>> > http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> extropy-chat mailing list
>>>>> extropy-chat at lists.extropy.org
>>>>> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>>>>>
>>>> _______________________________________________
>>>> extropy-chat mailing list
>>>> extropy-chat at lists.extropy.org
>>>> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>>>>
>>>>
>>>> _______________________________________________
>>>> extropy-chat mailing list
>>>> extropy-chat at lists.extropy.org
>>>> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>>>>
>>> _______________________________________________
>>> extropy-chat mailing list
>>> extropy-chat at lists.extropy.org
>>> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>>>
>>>
>>> _______________________________________________
>>> extropy-chat mailing list
>>> extropy-chat at lists.extropy.org
>>> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>>>
>> _______________________________________________
>> extropy-chat mailing list
>> extropy-chat at lists.extropy.org
>> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>>
>>
>> _______________________________________________
>> extropy-chat mailing list
>> extropy-chat at lists.extropy.org
>> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>>
> _______________________________________________
> extropy-chat mailing list
> extropy-chat at lists.extropy.org
> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20230226/f28a5737/attachment.htm>