[FRIAM] A* and emulatoin
Steve Smith
sasmyth at swcp.com
Mon Jun 27 14:30:05 EDT 2022
I accept your problems with "synthetic" as described and agree that
cobbled or mashed up has a more promising connotation than constructed
or designed or fabricated. It has been a *long* time since I thought
about that (faulty) dichotomy. The morality/ethics is also awkward for
me... I suppose it isn't as much about individual/group as
organic/engineered.
Since I've been reading Charlton *on Bateson* I am very reminded how
Bateson's seemly ideosyncratic language used to really put me off, but
now I feel more he was likely working in the interstices of meaning
between/among the conventional uses and landed on one bit of lexicon or
other awkwardly at least in part because the conventional use *was* off
in some way (or in it's precision it was naturally *wrong*?). I'm
still struggling/fumbling with it.
On 6/27/22 7:04 PM, glen wrote:
> Yeah, I don't like "synthetic" as much because it seems to rely on a
> false dichotomy between us and the other animals. Is a termite mound
> "synthetic"? Granted, "artificial" may hide some of that, too. But I
> think it's reasonable to say there are, say, naturally occurring
> (geological) mounds. Then there are artisan-generated, artificial,
> termite mounds, where the termites are the artisans. [⛧]
>
> And none of that artisanal stuff *requires* the artisan to
> reductionistically "understand" everything from first principles in
> the way "synthetic" might. "Synthetic" also often carries another
> false dichotomy between synthesis and analysis. It's false because
> nobody ever does pure [synthe|analy]sis. They're always done together.
> "Artificial" allows for that mode mixing. [We've had this discussion
> before in the usage of terms like "naturfact".]
>
> And that targets artificial morality nicely, I think. I've never
> really grokked the difference between morality and ethics, I think
> because making the distinction is a kind of composition/division
> fallacy. Ethics seems to carry the pretense of (or a slippery slope
> to) universality/monism, whereas morals seem to carry the pretense of
> individualism/relativism. If laid out on a spectrum, that's fine. But
> to draw a sharp line seems like sophistry.
>
> While I'm a consultant on a project regarding the ethics of AI in
> medicine, what interests me most is simulating the agency of an
> individual practitioner ... similar to the way we used to play
> red-blue-gray teams back at lockheed ... or the way you might simulate
> modern [cough] cyberwarfare.
>
>
> [⛧] Of course, you have to go all the way down to the 3rd defn in AH
> to find the right one. So if "synthetic" might mean "cobbled together
> from stuff you found lying around", then maybe it's better than
> "artificial". What I mean by both terms is closer to "glitch" ... a
> little bit of intent and a little bit of accident.
>
> AH "3. A phenomenon or feature not originally present or expected and
> caused by an interfering external agent, action, or process, as an
> unwanted feature in a microscopic specimen after fixation, in a
> digitally reproduced image, or in a digital audio recording."
>
> On 6/27/22 09:54, Steve Smith wrote:
>> I appreciate your addition of the 'M' to the *-match and want to
>> remind myself out loud in front of you that I once (and maybe should
>> again) preferred *synthetic* to *artificial*.... in the early days of
>> VR, "Artificial Reality" was in the running as a term, but I felt
>> *Synthetic Reality* carried the assertive sense of intentionality.
>> "Artificial" felt more passive... an artifact of a willful creation
>> with "Synthetic" feeling closer to the dynamic act of
>> *synthesizing*. And of course now (maybe not then), the spirit OF a
>> mashup vs a whole-cloth thing comes through with "Synthetic". This
>> of course before I came to learn the terms artifice and artificer in
>> this context.
>>
>> Is "Ethics" not in some sense *artificed* or *constructed*
>> morality? I don't know, it is definitely an interesting tangent to
>> all the other tangents that we tangent on here (tangentially). As
>> an aside, does a tangent of a tangent (of a tangent) imply higher and
>> higher derivatives, it seems like it is precisely that?! but in what
>> dimension?
>>
>> On 6/27/22 4:16 PM, glen wrote:
>>> Thanks very much for that link to mental contagion. It targets a
>>> number of problems I have with intersubjectivity, even if the
>>> author's nowhere near as skeptical as I think they should be. >8^D
>>>
>>> I drafted and deleted a response to Marcus' point about simple or
>>> high-order prediction. My draft targeted the distinction between
>>> [si|e]mulation more directly than yours. But yours homesteads a much
>>> more aggressive territory. (Tangentially, one of the A*'s I've been
>>> most interested in lately is AM - artificial morality. It turns out
>>> that simulation has a huge role to play in spoofing biases.)
>>>
>>> I intended to end that deleted post with my old rant about the (lack
>>> of a) difference between verification and validation ... a standard
>>> pedantic stance of gray bearded simulationists. I was once laughed
>>> out of the room at an SCS meeting for suggesting they're
>>> foundationally the same thing. Pffft!
>>>
>>> But all this hearkens back to the long-running thread on
>>> [in|ex]tensional attributes and the ontological status of their
>>> distinction. When is mimicry sufficient and when is "from whole
>>> cloth" necessary? As someone quipped re: Lemoine's attribution of
>>> sentience to LaMDA, "I have met meat Beings I consider less than
>>> sentient."
>>>
>>> On 6/25/22 23:55, Steve Smith wrote:
>>>> This is what made it through my semi-permeable filter-bubble
>>>> membrane first thing this morning (CET):
>>>>
>>>> https://theconversation.com/googles-powerful-ai-spotlights-a-human-cognitive-glitch-mistaking-fluent-speech-for-fluent-thought-185099
>>>>
>>>>
>>>> which became grist for the mill we have been grinding with here of
>>>> late. It highlights interesting things like how flawed (but
>>>> useful?) the Turing Test is. The TT represents precisely "the
>>>> glitch". I think this idea points in the general direction of
>>>> conscious empathy... if we recognize language fluency *as* mental
>>>> fluency, then it is more obvious that we would grant others who
>>>> present language fluency as being similar to ourselves, possibly
>>>> assuming that "other" is closer to "not other" simply because of
>>>> the familiar language that flows out of us.
>>>>
>>>> In my (limited) EU travels this season I have heard only a
>>>> half-dozen languages with half as many accents/dialects each... In
>>>> english-speaking ireland, a little gaelic slipped out here and
>>>> there but the accent referenced it with every lilt. This was not
>>>> unfamiliar to my ear, so I mostly heard it as "same", but in Wales,
>>>> the Welsh was not nearly (at all?) familiar and the
>>>> romanisation/anglification of the written Welsh was overwhelmingly
>>>> unfamiliar. When I read a sign, I felt like I was left with a
>>>> mouthful of consonants and diacritics that I had to spit out just
>>>> to clear my vocal passage to start on the next phrase.
>>>>
>>>> It gave me more sympathy for my non Southwest colleagues
>>>> struggling with the various anglifications of the hispanification
>>>> of a dozen different native American languages (starting in my
>>>> neighborhood with Tewa/Tiwa/Towa and expanding out withe Keres and
>>>> Dine' and Zuni ...) The (nearly conventional/normalized) rendering
>>>> of most of these languages is for me familiar enough that I don't
>>>> struggle or wince, but after (especially Welsh)... "I get it".
>>>> When confronted with each British accent (I couldn't identify or
>>>> distinguish many if any) it took a few hours at least to become
>>>> habituated enough to not be disturbed (intrigued or put off,
>>>> depending) by the unfamiliar sound patterns and often idiomatic
>>>> constructions.
>>>>
>>>> I thought i would be able to "hear" French as comfortably as I did
>>>> Italian 10 years ago, but it seems the "Romance" connections
>>>> between Spanish and Italian and the plethora of Latin words/phrases
>>>> in science made it much more familiar than French. The tiny bit of
>>>> French I think I am habituated to are a few Americanized stock
>>>> phrases and maybe a very little bit of dialogue from movies...
>>>> After a week of hearing almost nothing *but* French it no longer
>>>> felt outrageously "Other" even if I couldn't hardly parse a thing
>>>> out of a run-together-spoken-phrase. Mary and I observed one
>>>> another trying to speak English to someone who did not speak much
>>>> if any and we realized that we were both prone to repeat the same
>>>> sentence with a word choice or two changed, but more emphatically
>>>> (and therefore more run-together) each time. Not helpful, and
>>>> perhaps what the few French who bothered to speak to us once it was
>>>> established that we had no language in common, were doing
>>>> themselves. It was hard to recognize even word-breaks in the
>>>> word-salad coming at us. The little German we were exposed to
>>>> had a *different* set of familiar words and sounds and I think the
>>>> English and German might have a much stronger phonemic overlap,
>>>> making it not sound quite as foreign... though I was left wanting
>>>> to clear my throat after hearing much spoken german... and then
>>>> here in the Netherlands with *many*
>>>> English-speaking-with-Dutch-Accent we are much more comfortable...
>>>> and much of the written Dutch is familiar even when the
>>>> pronunciation is a git foreign.
>>>>
>>>> https://www.psychologicalscience.org/observer/the-cognitive-glitches-of-humans-laurie-santos-on-what-makes-the-human-mind-so-special
>>>>
>>>>
>>>> In trying to (re)find the first article, I ran across this article
>>>> which was a bit more interesting to me. The point they make about
>>>> human cognitive bias against anyone who speaks differently (acutely
>>>> illuminated by the once-familiar term "deaf and dumb" or
>>>> "dumb-mute" for those who could not speak (due to deafness,
>>>> aphasia, or perhaps some trauma? The line from the Rock Opera
>>>> "Tommy"s Pinball Wizard comes to mind: "That deaf, dumb and blind
>>>> kid, could sure play a mean pin ballll!"
>>>>
>>>> A counter to the *negative* bias I recently heard was: "Don't
>>>> mistake an accent for a personality"...
>>>>
>>>> It is fascinating to me how many ways we can split a hair in
>>>> discussing AI, etc. A* really. Intelligence, Reasoning, Life,
>>>> Consciousness, etc. ad nauseum. And yet it is useful (I think) to
>>>> note that no one of them is really broad nor narrow enough at the
>>>> same time. Each is a facet or reflection of the other. The second
>>>> article seems to discuss "emotional intelligence" or I think more
>>>> aptly "emotional knowledge". My very first (and practically
>>>> only) published "artpiece" was a visual study on the distinction
>>>> between "knowing" and "knowing-about", with AI climbing the steep
>>>> part of the hill toward a pinnacle (or more likely series of false
>>>> summits) of "knowing about" without possibly getting at all any
>>>> closer (at all) to "knowing".
>>>>
>>>> This leads me back to Marcus' haunting suggestion that "is learning
>>>> anything more than imitation/emulation?"
>>>>
>>>> Following Glen's ideation about bureaucracy as a form of tech, I
>>>> find that a great deal of my daily interaction with other people
>>>> is, in fact, with their bureaucratic roles. I am seeking a
>>>> transaction... knowledge, information, material goods, a service.
>>>> And given the level of the mutual (mis)understanding I've been
>>>> enduring for over a month now in those transactions, It now feels
>>>> like a luxury to expect a service person to articulate their
>>>> preferences and basis of their preferences in a given baked good,
>>>> bit of unfamiliar produce, or even (gawdess forbid) Beer! But it
>>>> has trained me to "listen for emotional content" more than
>>>> substance. If I ask for a "Blonde" or a "Bruun" or a "Trippel" or
>>>> a "Wit" and they rattle off something about one or more of them, I
>>>> will choose one based on the level of excitement in their voice-eye
>>>> over any imagined information content their response implied. I
>>>> am sometimes disappointed but almost always surprised. The
>>>> vocabulary of European Beers overlaps (up to language) what I am
>>>> familiar with amongst American Craft beers but my exploration is
>>>> wider (through clumsiness if nothing else). My best strategy is
>>>> simply to (try to) ask for "whatever is brewed locally". Also a
>>>> good strategy for food it seems.
>>>>
>
>
More information about the Friam
mailing list