Humanist Discussion Group, Vol. 40, No. 132.
        Department of Digital Humanities, University of Cologne
                      Hosted by DH-Cologne
                       www.dhhumanist.org
                Submit to: humanist@dhhumanist.org




        Date: 2026-08-26 09:58:51+00:00
        From: Tim Smithers <tim.smithers@cantab.net>
        Subject: Re: [Humanist] 40.123: "SO YOU SAY CHATBOTS ARE NARCISSITIC"

Dear Willard,

Thank you for the pointer to the Ciston, Berry et al book
about ELIZA. It's good to see this kind of careful and
thorough work published.

However, the last sentence from the words you quoted stood
out for me, because I so utterly disagree with them.

Preceding this last sentence we have ...

    "...  The history of ELIZA reminds us that any seemingly
     autonomous computational persona is ultimately a product
     of human design, user interpretation, and careful
     curation. ..."

Which looks good to me, but then, immediately following, we
get this contradictory claim ...

    "...  It becomes a collaborative performance that creates
     an illusion of an independent agent while obscuring the
     human hands that shape and maintain it."

Any and all interactions between real people and any kind of
Generative AI system is most certainly not a collaboration!
It is an interaction made by the system builders to look and
feel like a collaboration, perhaps.  But this is a deliberate
falsehood perpetrated by the builders and providers of these
systems.

'Collaboration' is, in my view, entirely the wrong word to use
here.  Collaboration, real collaboration, not the fake
collaboration we have talked about here, requires parties --
that's real people -- who have the needed thinking, reasoning,
and doing capacities, skills, knowledges, understandings, and
respect for their collaborators, to be able to effectively and
successfully work together on something.  Describing what goes
on between a person and any of today's "ChatBot" systems
severely damages our word and notion for what real
collaboration is, involves, and requires.

We need to stop doing this emptying out of the real meanings
of so many of our words, and their associated concepts, by
using them to describe and talk about what Generative AI
systems do when they don't.

We've now effectively lost, or are fast losing, all useful and
needed meanings of words like 'intelligence,' 'thinking,'
'reasoning,' 'knowing,' 'understanding,' 'learning,'
'writing,' 'reading,' 'summarising,' 'agent,' ...  the list
goes on and on.  The consequence is we now cannot make the
distinctions we need to make between how we do all these
things, and what Generative AI systems do and how they do it.
We seem to have learned nothing from Joseph Weizenbaum’s
strong warnings.  Including, disappointingly, Ciston, Berry,
et al, here.

Resisting this deliberate destruction of our languaging,
pushing back on it, recovering the needed real meanings of our
words for important things, is, surely, a job for Humanists to
lead, so that the rest of us may better see and understand
what's going on here, and grasp why we must all join in the
fight against it.

-- Tim




> On 17 Aug 2026, at 09:14, Humanist <humanist@dhhumanist.org> wrote:
>
>
>              Humanist Discussion Group, Vol. 40, No. 123.
>        Department of Digital Humanities, University of Cologne
>                      Hosted by DH-Cologne
>                       www.dhhumanist.org
>                Submit to: humanist@dhhumanist.org
>
>
>
>
>        Date: 2026-08-17 06:50:27+00:00
>        From: Willard McCarty <willard.mccarty@mccarty.org.uk>
>        Subject: chatbot narcissism
>
> Connection between chatbots used in the manner of ELIZA and Ovid's story
> of Narcissus in the Metamorphoses has been made numerous times. A recent
> alert from the Human Centered Artificial Intelligence people at Stanford
> gives us Zhang et al, "Interaction with AI Companions and Psychological
> Well-being" (4/5/2026, now in arXiv:2506.12605v5) to add to the list of
> relevant work along these lines, although it does not mention the
> Ovidian original. Likewise the very fine study by Ciston, Berry et al.,
> Inventing ELIZA: How the First Chatbot Shaped the Future of AI (MIT
> Press, 2026); see e.g. p. 134 there:
>
>> The legacy of ELIZA's personas can be said to live on in today's
>> ubiquitous Al assistants. But whereas Weizenbaum's hand-edited
>> transcripts made his editorial role visible to those who searched the
>> archives, the editorial choices made in training contemporary large
>> language models remain largely invisible to users. The history of
>> ELIZA reminds us that any seemingly autonomous computational persona
>> is ultimately a product of human design, user interpretation, and
>> careful curation. It becomes a collaborative performance that creates
>> an illusion of an independent agent while obscuring the human hands
>> that shape and maintain it.
>
> Someone (I forgot the source) has observed that turning to chatbots
> based on LLMs fixates our consciousness on the past, i.e. our historical
> selves beyond individual memory, as if the past were an oracle, not the
> future, not the Other, and so stunts our imaginations.
>
> Comments and further references to this line of thought would be most
> welcome.
>
> Best,
> WM
> --
> Willard McCarty,
> Professor em., King's College London;
> www.mccarty.org.uk



_______________________________________________
Unsubscribe at: http://dhhumanist.org/Restricted
List posts to: humanist@dhhumanist.org
List info and archives at at: http://dhhumanist.org
Listmember interface at: http://dhhumanist.org/Restricted/
Subscribe at: http://dhhumanist.org/membership_form.php