<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>Are LLMs nothing if not the epitome of  "gestalting" (aka "other
      ways of knowing")?</p>
    <p>for better and/or worse?<br>
    </p>
    <div class="moz-cite-prefix">On 9/15/24 2:08 PM, Marcus Daniels
      wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:MN0PR11MB598537596E4EA0A7CE32F675C5672@MN0PR11MB5985.namprd11.prod.outlook.com">
      <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
      <meta name="Generator" content="Microsoft Word 15 (filtered
        medium)">
      <style>@font-face
        {font-family:"Cambria Math";
        panose-1:2 4 5 3 5 4 6 3 2 4;}@font-face
        {font-family:Calibri;
        panose-1:2 15 5 2 2 2 4 3 2 4;}@font-face
        {font-family:Aptos;}p.MsoNormal, li.MsoNormal, div.MsoNormal
        {margin:0in;
        font-size:12.0pt;
        font-family:"Aptos",sans-serif;}a:link, span.MsoHyperlink
        {mso-style-priority:99;
        color:blue;
        text-decoration:underline;}span.EmailStyle19
        {mso-style-type:personal-reply;
        font-family:"Aptos",sans-serif;
        color:windowtext;}.MsoChpDefault
        {mso-style-type:export-only;
        font-size:10.0pt;
        mso-ligatures:none;}div.WordSection1
        {page:WordSection1;}</style><!--[if gte mso 9]><xml>
<o:shapedefaults v:ext="edit" spidmax="1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
<o:shapelayout v:ext="edit">
<o:idmap v:ext="edit" data="1" />
</o:shapelayout></xml><![endif]-->
      <div class="WordSection1">
        <p class="MsoNormal"><span style="font-size:11.0pt">How do we
            hold LLM (companies) accountable for their tool’s
            suggestions other showing their chain of reasoning and
            showing the evidence for the axioms they adopt?  That is, by
            adopting the scientific methods.  Can we expect LLMs to have
            “other ways of knowing”?   If not, why not?<o:p></o:p></span></p>
        <p class="MsoNormal"><span style="font-size:11.0pt"><o:p> </o:p></span></p>
        <div>
          <div style="border:none;border-top:solid #E1E1E1
            1.0pt;padding:3.0pt 0in 0in 0in">
            <p class="MsoNormal"><b><span
                  style="font-size:11.0pt;font-family:"Calibri",sans-serif">From:</span></b><span
style="font-size:11.0pt;font-family:"Calibri",sans-serif">
                Friam <a class="moz-txt-link-rfc2396E" href="mailto:friam-bounces@redfish.com"><friam-bounces@redfish.com></a> <b>On Behalf Of
                </b>Prof David West<br>
                <b>Sent:</b> Sunday, September 15, 2024 8:40 AM<br>
                <b>To:</b> <a class="moz-txt-link-abbreviated" href="mailto:friam@redfish.com">friam@redfish.com</a><br>
                <b>Subject:</b> Re: [FRIAM] affinity for chatbots<o:p></o:p></span></p>
          </div>
        </div>
        <p class="MsoNormal"><o:p> </o:p></p>
        <div>
          <p class="MsoNormal"><span
              style="font-family:"Arial",sans-serif">Roger,<o:p></o:p></span></p>
        </div>
        <div>
          <p class="MsoNormal"><span
              style="font-family:"Arial",sans-serif"><o:p> </o:p></span></p>
        </div>
        <div>
          <p class="MsoNormal"><span
              style="font-family:"Arial",sans-serif">I love
              this post. Although NOT what you intended, I find it a
              scathing (if a bit indirect) indictment of scientism (the
              privileging of the scientific method) and of Pierce's
              truth as reasoned consensus philosophy.<o:p></o:p></span></p>
        </div>
        <div>
          <p class="MsoNormal"><span
              style="font-family:"Arial",sans-serif"><o:p> </o:p></span></p>
        </div>
        <div>
          <p class="MsoNormal"><span
              style="font-family:"Arial",sans-serif">i can
              only hope to meet an unbiased LLM. Maybe as entertaining
              and enlightening as my conversations with fellow acid
              heads.<o:p></o:p></span></p>
        </div>
        <div>
          <p class="MsoNormal"><span
              style="font-family:"Arial",sans-serif"><o:p> </o:p></span></p>
        </div>
        <div>
          <p class="MsoNormal"><span
              style="font-family:"Arial",sans-serif">davew<o:p></o:p></span></p>
        </div>
        <div>
          <p class="MsoNormal"><span
              style="font-family:"Arial",sans-serif"><o:p> </o:p></span></p>
        </div>
        <div>
          <p class="MsoNormal"><span
              style="font-family:"Arial",sans-serif"><o:p> </o:p></span></p>
        </div>
        <div>
          <p class="MsoNormal">On Sun, Sep 15, 2024, at 8:06 AM, Roger
            Critchlow wrote:<o:p></o:p></p>
        </div>
        <blockquote style="margin-top:5.0pt;margin-bottom:5.0pt" id="qt">
          <div>
            <div>
              <p class="MsoNormal">The Agile versus Waterfall contrast
                sounds like a variation of Exploration versus
                Exploitation.  I'm glad nuclear decommissioning isn't
                running Reinforcement Learning, that could lead to some
                very unfortunate explorations.<o:p></o:p></p>
            </div>
            <div>
              <p class="MsoNormal"><o:p> </o:p></p>
            </div>
            <div>
              <p class="MsoNormal">It's odd to hear Residual Bias spoken
                of as something that should eventually go away, when it
                seems like it's here to stay, the original sin of
                language, never to be expunged from the LLM's until they
                renounce language entirely.  That is, language is a
                collective behavior based on the sharing of individual
                experiences, hence it's hostage to the set of
                experiences which actually happened or were imagined to
                happen, and to the subset of those which were shared,
                however that turned out.  So it all starts with a bias
                against experiences which people didn't have, didn't
                imagine, didn't share, failed to communicate, or
                forgot.  We have no idea what's in that set of excluded
                stuff or how big it is.  When we build LLM's we add
                another bias against those expressions of language which
                are not in the training set.  Then we censor the models,
                adding another layer of bias to remove ugliness.  Then
                we talk about the Residual Bias as if all of this could
                be portrayed as some principled approach to perfection
                and we're measuring the goodness of fit.  So if Dave
                thinks the uncensored LLM's were wild and crazy, wait
                until he meets an unbiased LLM.<o:p></o:p></p>
            </div>
            <div>
              <p class="MsoNormal"><o:p> </o:p></p>
            </div>
            <div>
              <p class="MsoNormal">-- rec --<o:p></o:p></p>
            </div>
          </div>
          <div>
            <p class="MsoNormal"><o:p> </o:p></p>
          </div>
          <div>
            <div>
              <p class="MsoNormal">On Sat, Sep 14, 2024 at 11:51<span
                  style="font-family:"Arial",sans-serif"> </span>PM
                glen <<a href="mailto:gepropella@gmail.com"
                  moz-do-not-send="true" class="moz-txt-link-freetext">gepropella@gmail.com</a>>
                wrote:<o:p></o:p></p>
            </div>
            <blockquote style="border:none;border-left:solid #CCCCCC
              1.0pt;padding:0in 0in 0in
              6.0pt;margin-left:4.8pt;margin-right:0in">
              <div>
                <p class="MsoNormal">Both Roger's and Marcus' replies
                  mentioned the co-construction of *the* world, at least
                  indirectly. Your concept of narrowing sounds to me
                  like a refining, rather than a narrowing. In order to
                  refine, you do have to narrow the scope (or decrease
                  the focal length of your lens), but you're not
                  narrowing the world. I'd argue you're enlarging the
                  world by adding detail in a "dense" way ... in the
                  interstitial spaces between coarse constraints.<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal"><o:p> </o:p></p>
              </div>
              <div>
                <p class="MsoNormal">One possible flaw in both Roger's
                  (or Irene's?) argument that the act of explanation
                  facilitates understanding is, from a pluralist
                  perspective, if we really are co-constructing the
                  world, then such exercises in explaining are simply
                  narrative-reinforcers. The chatbots are good at
                  telling stories, but less good at teaching the core
                  curiosity necessary for having experiences from which
                  stories can be told ... story-generators are different
                  from story-repeaters ... I guess it's like the old
                  distinction between teaching and doing. Sabine's
                  admiration of flat earthers is good, if awkward, along
                  these lines: <a
                    href="https://youtu.be/f8DQSM-b2cc?si=xyqpS2FJjH4imOy4"
                    target="_blank" moz-do-not-send="true"
                    class="moz-txt-link-freetext">https://youtu.be/f8DQSM-b2cc?si=xyqpS2FJjH4imOy4</a><o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal"><o:p> </o:p></p>
              </div>
              <div>
                <p class="MsoNormal">That has consequences to your sense
                  of the chatbot pushing you toward homogeny and a risk
                  in Marcus' abdicating to the chatbots, as well.
                  Unnecessary anecdote: I was just discussing the role
                  SpaceX has played in demonstrating Agile versus
                  Waterfall approaches with a nuclear decomissioning
                  consultant (yes, at the pub, of course). Given her
                  role(s), she's naturally more inclined to the latter.
                  Having a good conception of the end-of-life status for
                  something like nuclear power requires significant
                  look-ahead. And I'm far from an Elno advocate. But
                  there's a kind of meta-processing we have to go
                  through in deciding where Agile is best versus where
                  Waterfall is best. I sincerely doubt either of us
                  could have had such an argument with a chatbot, even
                  in the medium-flung future.<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal"><o:p> </o:p></p>
              </div>
              <div>
                <p class="MsoNormal">On 9/13/24 11:34, steve smith
                  wrote:<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> Glen -<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> I appreciate your speaking
                  more directly to these thoughts/ideas than we have
                  been here.   I have been moved by your assertions
                  about vocal (linguistic?) grooming since you first
                  introduced them.   I am recently finished reading
                  Sopolsky's "Primate's Memoir" which adds another
                  dimension/parallax-angle (for me) on intertribal
                  behaviour among primates beyond the more familiar
                  Chimpanzee and of late Bonobo.<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> I am just now also just
                  finishing (re-reading parts) of Kara Swisher's "Burn
                  Book" which covers her own experience/perspective
                  across TechBro culture where a pretty significant
                  amount of Alpha/Beta pecking order exhibits itself and
                  we see the current rallying of (too) much of that sub
                  culture to MAGA/Trump fealty.<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">>> We've talked about how
                  some of us really enjoy simulated conversation with
                  chatbots ... "really" is an understatement ... it
                  looks more like a fetish or a kink to me ... too
                  intense to be well-described as "enjoyment". Anyway,
                  this article lands in that space, I think:<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> I will confess to having an
                  "appreciation" for the "simulated conversation to
                  which you refer... It might have reached kink or
                  fetish levels for a little while when I was first
                  exploring the full range of GPT 3.5 and then 4.0
                  available to me.  I've referred to GPT as my "new bar
                  friend" or maybe to the point a little like finding a
                  new watering hole with a number of regulars who I can
                  find a qualitatively new conversation.<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> I've mostly moved past that
                  fascination...  I'm not as surprised by these "new
                  friends" as I was for the first few months of dropping
                  in on them.<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">>> It seems to me that some
                  arbitrary thought can play at least a few roles to a
                  person. It may provide: 1) a kernel of identity to
                  establish us vs. them, 2) fodder for feigning
                  engagement at cocktail parties and such, and 3) a foil
                  for world-construction (collaboratively or
                  individually).<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">>><o:p> </o:p></p>
              </div>
              <div>
                <p class="MsoNormal">>> (1) and (2) wouldn't
                  necessarily mechanize refinement of the thought,
                  including testing, falsification, etc. But (3) would.
                  For me, (2) does sometimes provide an externalized
                  medium by which I can change my mind. Hence my
                  affinity for argument, especially with randos at the
                  pub. But it seems like coping and defense mechanisms
                  like mansplaining allow others to avoid changing their
                  minds with (2).<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> Like you (only very
                  differently in detail I am sure) I tend to push my
                  chatbot "friends" until they begin to contradict me or
                  argue with me. While some of the discussions involve
                  "worldbuilding"  I think of it more as "world
                  narrowing"?   In my case meaning, helping me think and
                  talk my way through a *subset* of the possibilities I
                  see on "solving a problem" which might be more
                  appropriately framed as building a problem-space world
                  and then narrowing (or even bending) the solution
                  space away from the conventional.<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> For example discussing (at
                  excruciating length) the design and construction of a
                  modest addition on my home,  starting with fairly
                  conventional big-box-available industrial solutions
                  but evolving toward using locally sourced, somewhat
                  more natural materials (soilcrete, rough-sawn timbers
                  from nearby, scoria/perlite for in-ground insulation,
                  mycelium (grown in loose cellulose, oat-straw or
                  hemp-fibers) roof and wall insulation, etc.  Most of
                  my DIY friends are capable of engaging in this but
                  their idiosyncratic (as opposed to my own) preferences
                  (fetishes and fears) tend to taint the dialog a
                  little.  GPT *does* try to channel me back to the
                  conventional, offering reasons why I really *should*
                  consider using the most conventional
                  materials/methods. Nevertheless if I speak in
                  reasonable and coaxing tones it will usually
                  acknowledge that their are contexts wherein my ideas
                  might be viable (though there always remains a
                  skeptical bias) and in fact helps me split hairs on
                  just <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> what might be the contexts
                  where my ideas *are* viable...<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">>><o:p> </o:p></p>
              </div>
              <div>
                <p class="MsoNormal">>> Another concept I've
                  defended on this list is the vocal grooming
                  hypothesis. If a lonely person engages a chatbot as a
                  simple analogy to picking lice from others' fur, then
                  their engagement with the bot probably lands squarely
                  in (1) and (2). But if the person is simply an
                  introverted hermit who has trouble co-constructing the
                  world with others (i.e. *not* merely vocal grooming),
                  then the chatbot does real work, allowing the
                  antisocial misfit to do real work that could later be
                  expressed in a form harvestable by others. I wonder
                  what humanity could have harvested if Kaczynski or
                  Grothendieck in his later years had had access to
                  appropriately tuned chatbots.<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> I'd like to think the chatbots
                  I hang out with might have helped them talk themselves
                  *out* of their most acute anti-social activities...
                  but maybe not.<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">> <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal"><o:p> </o:p></p>
              </div>
              <div>
                <p class="MsoNormal"><o:p> </o:p></p>
              </div>
              <div>
                <p class="MsoNormal">-- <o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">glen<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal"><o:p> </o:p></p>
              </div>
              <div>
                <p class="MsoNormal">-. --- - / ...- .- .-.. .. -.. / --
                  --- .-. ... . / -.-. --- -.. .<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">FRIAM Applied Complexity Group
                  listserv<o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">Fridays 9a-12p Friday St. Johns
                  Cafe   /   Thursdays 9a-12p Zoom <a
                    href="https://bit.ly/virtualfriam" target="_blank"
                    moz-do-not-send="true" class="moz-txt-link-freetext">https://bit.ly/virtualfriam</a><o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">to (un)subscribe <a
                    href="http://redfish.com/mailman/listinfo/friam_redfish.com"
                    target="_blank" moz-do-not-send="true"
                    class="moz-txt-link-freetext">http://redfish.com/mailman/listinfo/friam_redfish.com</a><o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">FRIAM-COMIC <a
                    href="http://friam-comic.blogspot.com/"
                    target="_blank" moz-do-not-send="true"
                    class="moz-txt-link-freetext">http://friam-comic.blogspot.com/</a><o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">archives:  5/2017 thru present <a
href="https://redfish.com/pipermail/friam_redfish.com/" target="_blank"
                    moz-do-not-send="true" class="moz-txt-link-freetext">https://redfish.com/pipermail/friam_redfish.com/</a><o:p></o:p></p>
              </div>
              <div>
                <p class="MsoNormal">  1/2003 thru 6/2021  <a
                    href="http://friam.383.s1.nabble.com/"
                    target="_blank" moz-do-not-send="true"
                    class="moz-txt-link-freetext">http://friam.383.s1.nabble.com/</a><o:p></o:p></p>
              </div>
            </blockquote>
          </div>
          <div>
            <p class="MsoNormal">-. --- - / ...- .- .-.. .. -.. / -- ---
              .-. ... . / -.-. --- -.. .<o:p></o:p></p>
          </div>
          <div>
            <p class="MsoNormal">FRIAM Applied Complexity Group listserv<o:p></o:p></p>
          </div>
          <div>
            <p class="MsoNormal">Fridays 9a-12p Friday St. Johns Cafe  
              /   Thursdays 9a-12p Zoom <a
                href="https://bit.ly/virtualfriam"
                moz-do-not-send="true" class="moz-txt-link-freetext">https://bit.ly/virtualfriam</a><o:p></o:p></p>
          </div>
          <div>
            <p class="MsoNormal">to (un)subscribe <a
                href="http://redfish.com/mailman/listinfo/friam_redfish.com"
                moz-do-not-send="true" class="moz-txt-link-freetext">http://redfish.com/mailman/listinfo/friam_redfish.com</a><o:p></o:p></p>
          </div>
          <div>
            <p class="MsoNormal">FRIAM-COMIC <a
                href="http://friam-comic.blogspot.com/"
                moz-do-not-send="true" class="moz-txt-link-freetext">http://friam-comic.blogspot.com/</a><o:p></o:p></p>
          </div>
          <div>
            <p class="MsoNormal">archives:  5/2017 thru present <a
                href="https://redfish.com/pipermail/friam_redfish.com/"
                moz-do-not-send="true" class="moz-txt-link-freetext">https://redfish.com/pipermail/friam_redfish.com/</a><o:p></o:p></p>
          </div>
          <div>
            <p class="MsoNormal">  1/2003 thru 6/2021  <a
                href="http://friam.383.s1.nabble.com/"
                moz-do-not-send="true" class="moz-txt-link-freetext">http://friam.383.s1.nabble.com/</a><o:p></o:p></p>
          </div>
          <div>
            <p class="MsoNormal"><o:p> </o:p></p>
          </div>
        </blockquote>
        <div>
          <p class="MsoNormal"><span
              style="font-family:"Arial",sans-serif"><o:p> </o:p></span></p>
        </div>
      </div>
      <br>
      <fieldset class="moz-mime-attachment-header"></fieldset>
      <pre class="moz-quote-pre" wrap="">-. --- - / ...- .- .-.. .. -.. / -- --- .-. ... . / -.-. --- -.. .
FRIAM Applied Complexity Group listserv
Fridays 9a-12p Friday St. Johns Cafe   /   Thursdays 9a-12p Zoom <a class="moz-txt-link-freetext" href="https://bit.ly/virtualfriam">https://bit.ly/virtualfriam</a>
to (un)subscribe <a class="moz-txt-link-freetext" href="http://redfish.com/mailman/listinfo/friam_redfish.com">http://redfish.com/mailman/listinfo/friam_redfish.com</a>
FRIAM-COMIC <a class="moz-txt-link-freetext" href="http://friam-comic.blogspot.com/">http://friam-comic.blogspot.com/</a>
archives:  5/2017 thru present <a class="moz-txt-link-freetext" href="https://redfish.com/pipermail/friam_redfish.com/">https://redfish.com/pipermail/friam_redfish.com/</a>
  1/2003 thru 6/2021  <a class="moz-txt-link-freetext" href="http://friam.383.s1.nabble.com/">http://friam.383.s1.nabble.com/</a>
</pre>
    </blockquote>
  </body>
</html>