Humanist Discussion Group

Humanist Archives: March 21, 2023, 7:50 a.m. Humanist 36.463 - followup: agency & intelligence

              Humanist Discussion Group, Vol. 36, No. 463.
        Department of Digital Humanities, University of Cologne
                      Hosted by DH-Cologne
                Submit to:

    [1]    From: Tim Smithers <>
           Subject: Re: [Humanist] 36.461: followup: agency & intelligence (120)

    [2]    From: maurizio lana <>
           Subject: Re: [Humanist] 36.461: followup: agency & intelligence (47)

    [3]    From: Willard McCarty <>
           Subject: books not absolutely dead (28)

        Date: 2023-03-20 17:11:35+00:00
        From: Tim Smithers <>
        Subject: Re: [Humanist] 36.461: followup: agency & intelligence


Okay, James, let's understand your line on anthropomorphism in
text as a leaking of the author's agency in to the text they
write, to be read by the agency of a reader.  That makes the
text feel pushy, perhaps a bit arrogant.  Acceptable in
places, yes.  Useful sometimes.  Needed occasionally.
Dramatically effective, when done with artful skill.  But
still inappropriate in places such as research communications,
I think, where, too often, it's used as a false, or mistaken,
façade for a claimed, or imagined, "objectivity."

Now, what happens when the producer of the text has no
intentional agency?  Yes, I want to introduce this needed
aspect of agency; needed for a useful notion of agency, rather
than the cheap notion we use to give texts agency, and to say
things like the pebbles on the beach act to stay in their
places on the beach ...  Except, of course, they don't stay in
their places.  The next tide comes in and moves them.  So the
tide has agency?  It does move the pebbles on the beach, but
it doesn't have any intention to do so, I would say, though
King Cnut may have said different.

ChatGPT has no intentional agency.  It produces text that a
human reader may read intention into, yes, but nowhere in all
ChatGPT's innards will any source or mechanisms of this
intention be found.  A human reader may "feel" an intention is
clearly there, expressed through the text, yes, but this
reader's agency does not, and cannot, put intentional agency
into ChatGPT, nor, of course, in to any human author either.

So, I would argue, we should inspect a text for the
intentional agency behind it, and any that has perhaps been
allowed to leak in to it, deliberately, or thoughtlessly.  You
see, I hope, there is a difference between cheap agency and
intentional agency, and it's this difference, I think, that
makes all text generate by ChatGPT, and its ilk, Plastic Text,
with no real intentional agency behind it, and not like human
made text, with real intentional agency behind it, and
sometimes in it, like you cite some examples of.  When ChatGPT
writes "I" there is no real I. When ChatGPT writes "This
paragraph explains ...", or some such anthropomorphism, there
is no author's intentional agency leaking in here.

It's the same for generative systems that produce images, or
sounds.  These other systems produce only Plastic Images and
only Plastic Sounds.  It's us humans who will decide if we
will fill our lives with Plastic Texts, Plastic Images,
Plastic Sounds, etc, to go with our Plastic Flowers, or prefer
real flowers, and real text, real images, real sounds.

I'm off to join Ferdinand.

Best regards,


> On 20 Mar 2023, at 06:02, Humanist <> wrote:
>              Humanist Discussion Group, Vol. 36, No. 461.
>        Department of Digital Humanities, University of Cologne
>                      Hosted by DH-Cologne
>                Submit to:
>        Date: 2023-03-18 13:32:43+00:00
>        From: James Rovira <>
>        Subject: Re: [Humanist] 36.457: followup: agency & intelligence
> Many thanks to Ann James and Tim Smithers for their responses. Tim's
> response reminds me of a review editor who very recently asked me to revise
> my first sentence to ascribe agency to the author, not the book -- which
> reminds me that I ask undergrads to go ahead and ascribe agency to written
> texts rather than start every other sentence with, "In 'Title of
> Work',"....". But that phrase is always followed by the author's name. This
> concept of the agency of written texts apart from direct authorial
> intervention is an old one. Milton has a poem, for example, describing
> himself as a king and his poem as his ambassador acting on his behalf,
> independently of him, with no small expression of anxiety. Ann Bradstreet's
> "The Author to Her Book," close in period to Milton, uses a similar trope,
> except her book is her wayward child rather than her ambassador.
> So I think we have a workable paradigm here, seriously, that we could
> extend to large language models. It bugs me that I didn't think of it
> before. I think we should keep in mind, though, that the text only has
> indirect or unintentional agency -- direct agency is in the author and then
> again in the reader's act of interpretation at the other end, so that the
> author's anxiety is not about how well the text actively represents him,
> but what his readers are going to do with it. We see these ideas repeated
> in Death of the Author paradigms that really appear early 20thC with T.S.
> Eliot and then reappear in structuralism and post-structuralism.
> Tim needs to follow some of my friends on FB, who are having a great deal
> of fun with ChatGPT. One asked for a version of Who's on First using Star
> Trek characters. The program, apparently, couldn't distinguish between
> descriptions of how the humor works and an actual representation of the
> humor. They've been silent about it a couple of weeks, though, so I think
> they've become bored and moved on.
> Much appreciation for Ann's recommendations. Agency is such a big topic
> that we could look for models in a variety of sources and approaches.
> Jim R
> --
> Dr. James Rovira <>
>   - *David Bowie and Romanticism
>   <>*,
>   Palgrave Macmillan, 2022
>   - *Women in Rock, Women in Romanticism
>   <
> Emancipation-of-Female-Will/Rovira/p/book/9781032069845>*,
>   Routledge, 2023

        Date: 2023-03-20 08:42:10+00:00
        From: maurizio lana <>
        Subject: Re: [Humanist] 36.461: followup: agency & intelligence

the discussion is very interesting.

i would like to recall that the supposed agency of the book is nothing
but the well known trope of "naming the work in place of the author". we
use it currently for songs, paintings, etc (the songs says that...). so
i would be cautious before making a special case for the fact that we
see it used also for the IA systems - even if for the songs no one
ignores the role of the author (sometimes split between the person who
sings and the person who wrote the words) while for the IA systems many
ignore the role of the authors. and this makes a difference.
i would also like to recall that the very fact that behind any IA system
there is an author (there are many authors), is what moves the EU
legislative action see for example attributing to the producers (=the
authors; plus the backend operators!) the civil liability for the
consequences of the use of IA systems. see the European Parliament
resolution of 20 October 2020 with recommendations to the Commission on
a civil liability regime for artificial intelligence,
> The European Parliament ... considers that the Product Liability
> Directive (PLD) has, for over 30 years, proven to be an effective
> means of getting compensation for harm triggered by a defective
> product, [...]; urges the Commission to assess whether the PLD should
> be transformed into a regulation, to clarify the definition of
> ‘products’ by determining whether digital content and digital services
> fall under its scope and to consider adapting concepts such as
> ‘damage’, ‘defect’ and ‘producer’; is of the opinion that, for the
> purpose of legal certainty throughout the Union, following the review
> of the PLD, the concept of ‘producer’ should incorporate
> manufacturers, developers, programmers, service providers as well as
> backend operators;
under the formal language there is the concept that any expression of
the type "the IA system did..." has the juridical meaning of "the
producers and the the backend operators of the IA system did..."


il pubblico uso della propria ragione deve sempre essere libero
immanuel kant

Maurizio Lana
Università del Piemonte Orientale
Dipartimento di Studi Umanistici
Piazza Roma 36 - 13100 Vercelli

        Date: 2023-03-20 05:32:01+00:00
        From: Willard McCarty <>
        Subject: books not absolutely dead

Thanks to Jim Rovira for reminding us of Milton's Areopagitica: "for
books are not absolutely dead things, but do contain a potency of life
in them to be as active as that soul whose progeny they are; nay, they
do preserve as in a vial the purest efficacy and extraction of that
living intellect that bred them". Some, such as Ong, point out that
books are here not said to be alive, but I think it takes the reader to
see to that--and what life can arise from the right book at the right
time. Just as (to cite Mircea Eliade) ancient burial rites provided
stone bodies for the dead so that they could be strong like rocks, so
libraries for books.

There's something here that interests me very much--and has to do 
with what we call 'arrtificial intelligence'. I think perhaps we can get 
there (to an understanding of this interesting thing) perhaps by asking 
'where does this coming alive happen?' Is the book like a catalyst? Or, 
when we have a conversation that really gets somewhere, where is 
this 'somewhere'? Do we not create problems for ourselves by 
positing or assuming an absolute separation of things, by an 
insistence on objectivity?


Willard McCarty,
Professor emeritus, King's College London;
Editor, Interdisciplinary Science Reviews;  Humanist

Unsubscribe at:
List posts to:
List info and archives at at:
Listmember interface at:
Subscribe at: