Humanist Discussion Group, Vol. 37, No. 89.
Department of Digital Humanities, University of Cologne
Hosted by DH-Cologne
www.dhhumanist.org
Submit to: humanist@dhhumanist.org
Date: 2023-06-08 09:42:15+00:00
From: Tim Smithers <tim.smithers@cantab.net>
Subject: Re: [Humanist] 37.78: a comment on an old question
Dear Manfred,
I appreciated your response to Willard's two questions.
From this, your question
"... how DO you answer the question what something is,
which you cannot identify with?"
stood out.
For me, it's not Digital Humanities I cannot identify with, I
don't really, it's Artificial Intelligence, AI, which I used
to identify with, and still try to. But my difficulty isn't
'cos AI has splurged into all sorts of AI-y things, it's
because it seems to have become just one thing, so called Deep
Learning, and I don't think this has much to do with AI. To
me, so called machine learning is a way of programming
computers to do certain kinds of things. It doesn't help us
understand more, or better, human (or other animal) learning.
(And calling it learning just empties out a useful term of
some needed meaning.)
The term Artificial Intelligence, for me, is the name of a
field of research, like Computer Science, CS, is. Unlike for
some, I think AI is not a part of CS. AI is a close neighbour
of CS, but not well understood as being a topic in CS. This
reflects where I learned most of my AI, in the Department of
AI (as it was then called) at Edinburgh University, which, at
about the same time, also set up a separate Department of
Computer Science.
To me, an AI is not a thing; it's not some kind of human made
machine. So when, just to take one of many suitable examples,
Yuval Noah Harari argues that "AI has hacked the operating
system of human civilisation" [TE2023.04.30], what I read
Harari as saying is that the whole community of AI researchers
has hacked the operating system of human civilisation, which
certainly does sound dramatic, and like a good cause to be
worried. What Harari doesn't explain is how we AI researchers
were allowed to get away with this massive hack.
But, Harari is, of course, not using the term AI to refer to
the field of research. AI, here, like in loads of other
expressions of fear and worry we see these days, refers to a
thing, an artificially intelligent thing, and a thing so
intelligent we should be worried by it, a thing so intelligent
it has decided to hack human civilisation.
This is a case of reification gone wild, and, I would say, a
rather silly case, not to say stupid. (This is not the first
time Artificial Intelligence has met with Human Stupidity.)
Intelligence is a human concept, not a property, quality,
aspect, capacity, of humans and other animals, though
sometimes, I admit, we do talk, loosely, of intelligence as if
it is one of these. We don't have a well established, widely
accepted, characterisation, or definition, of something we can
all happily agree to call intelligence. The reason for this
is, I think, because we don't need such a definition. The
concept of intelligence does plenty of useful work for us as a
label for a concept: a usefully stretchy and flexible concept,
like many of our concepts are. In this way, it does, or,
better said, did, useful work in the name Artificial
Intelligence for a field of research.
When I arrived in the AI Department in Edinburgh in 1984, from
being a structural designer and research in computational
techniques for structural analysis, including work on parallel
algorithms and parallel hardware -- to make the analysis go
faster, using the Inmos Transputer, for those who remember
those -- what I found was a collection of different research
groups each working on some particular topic: knowledge
representation and inference, natural language processing,
mathematical reasoning, machine vision, logic programming and
its implementation (as Prolog), AI in Education (using Logo),
Artificial Neural Networks, AI in Design, and Intelligent
Robotics (these last two being what I did). Nobody worked on
intelligence. Nobody said they were investigating
intelligence. And this is how it was in all the other AI
Departments and AI Research Labs I came to know through
various collaborations and visits. The collection of topics
being worked on in these different places varied some, and the
ways people did their research was different, but "what is
intelligence" was never part of any serious conversation or
discussion, or research. We talked about how we were trying
to get some kind of digital computation, programmed in some
particular way, or ways, to do some particular kind of thing:
represent and usefully reason about some aspect of what goes
in some real designing, for example.
I came to like AI as a field. I found many different kinds of
people from a greater range of backgrounds and interested than
I had found in the large Engineering Department where I did my
PhDing. And, I found people more easily able to talk to each
other, despite working on quite different stuff. When AI
research got bashed from the outside, which it sometimes did,
we came together to defend and support the particular kind of
research taking the hit. And, when some of us got too
excited, which we sometimes did, and made unwise over-the-top
claims and predictions about what would soon be achieved, we
collectively advised greater caution. And, we made some
progress on some particular things, useful things, as well as
discovered what we didn't understand well, and were not able
to make happen in the artificial.
It stayed this way until about ten years ago, give or take
some, depending on where you were. I still know AI people I
have worked with at some time, but it doesn't feel like we are
part of a community of AI researchers. Work in AI now seems
to be almost mono-topic, so called machine learning, or,
because that doesn't sound grand enough, deep learning, but
all using so called artificial neural network techniques --
which, of course, bear almost no real resemblance to what we
currently understand about how real neural systems work. Yes,
serious, and important advances have been made, but, with
these seem to have come more daft, not to say stupid, claims
about what these advances actually deliver, Artificial General
Intelligence (AGI), for example.
To keep healthy, the community of people working in some named
field of research and scholarship should, I think, sustain
some conversation about what they, as a community, are doing,
how they are doing it, what contributions have been, and are
being, made, to whom, and, occasionally, what they want the
name of their field of work to mean. I think we used to do
this in AI, but not any more. And I think this has made doing
real AI research a lot harder, perhaps even impossible. My
prediction for all this deep learning, generative AI, large
language models, stuff is, once the shouting has stopped and
the smoke has cleared -- which may take a while -- things like
ChatGPT will be regarded like we think of ball-point-pens
today: a useful convenience. I remember lots of shouting and
scare stories when ball-point-pens were first introduced in
schools in the UK. They have had a lasting effect, yes, but
they have not transformed nor destroyed the world as we know
it, which is what some people said would happen.
Still, I don't see a return to sensible AI (as a field of
research) happening any time soon. From this perspective,
Digital Humanities (DH) looks to me to be in quite a good
state of health. Talking about how and why it is, and what
counts as DH, is, however, I think, a good thing to be doing.
Best regards,
Tim
[TE2023.04.30] : Yuval Noah Harari argues that AI has hacked
the operating system of human civilisation,
The Economist, 30 May, 2023, by invitation.
> On 6 Jun 2023, at 07:26, Humanist <humanist@dhhumanist.org> wrote:
>
>
> Humanist Discussion Group, Vol. 37, No. 78.
> Department of Digital Humanities, University of Cologne
> Hosted by DH-Cologne
> www.dhhumanist.org
> Submit to: humanist@dhhumanist.org
>
>
>
>
> Date: 2023-06-05 14:55:49+00:00
> From: Manfred Thaller <manfred.thaller@uni-koeln.de>
> Subject: Re: [Humanist] 37.15: two questions
>
> Dear Willard,
>
> it is some time since you asked these questions, having been more or
> less incommunicado for some time, I'd nevertheless try to not so much
> answer, but comment on one of them.
>
> To recapitulate, you wrote:
>
>> But still one of them asked me, "So... what IS digital
>> humanities?" I was not lost for words, ...
> ...
>> Ruminating on the incident this morning has led me to wonder how others
>> answer that question in brief. So, in a nutshell, what is it? Does anyone
>> have a good one or two-sentence response?
> No, I have NO good one or two sentence answer. But I'd like to
> speculate, where one could go to find one.
>
> Reacting to the same question Tim Smithers wrote:
>> people using computation to do the kinds of research
>> scholars in the Humanities do.
>
> Well, around the year 2000 I'd unreservedly endorsed that answer, and
> with some growing reluctance I even used it myself in certain
> rhetorical situations until ca. 2015, becoming more and more
> dissatisfied with it, as I increasingly encountered items labeled "DH"
> where I found it difficult to discover the "Digital", the "Humanities"
> or both.
>
> Then the scales fell from my eyes when I read "Alternative
> Historiographies of the Digital Humanities" edited by Dorothy Kim and
> Adeline Koh, punctum books, 2021. There Dorothy Kim writes in the
> editor's introduction on p. 24:
>> Alternative Historiographies of the Digital Humanities resists
>> a linear history of the digital humanities — a straight line from
>> the beginnings of humanities computing. By discussing alterna-
>> tives histories of the digital humanities that address queer gam-
>> ing; feminist game studies praxis; Cold War military-industrial
>> complex computation; the creation of the environmental hu-
>> manities; monolingual discontent in DH; the hidden history of
>> DH in English studies; radical media praxis; cultural studies and
>> DH; indigenous futurities; Pacific Rim postcolonial DH; the issue
>> of scale and DH; Black feminist praxis; Global African feminist
>> protest; Black feminist archives; and the racialized silences in
>> topic modeling; the radical, indigenous, feminist histories of the
>> digital database; and the possibilities for an antifascist DH, this
>> collection hopes to re-set discussions of the DH and its attend-
>> ing straight, white origin myths.
>
> This really was true enlightenment. All my confusion, why at events
> labeled DH I had to listen to panels discussing neither the Humanities
> nor anything related to computers was suddenly resolved. The Digital
> Humanities are all that which Kim & Koh list, I am happy to agree.
>
> It's just that I am not particularly interested in most of it.
>
> But that does not answer your question - how DO you answer the question
> what something is, which you cannot identify with?
>
> I have no real two sentence answer, but when I am asked a similar
> question, I usually use a formula like: "Well, you know, that is a very
> wide field. MY personal interest in it are all things
> <fill-in-personal-definition;-Tim's-qualifying>."
>
> If I am not directly asked, I do not use the term - or use it in "the so
> called Digital Humanities".
>
> Tim, by the way:
>> What if, instead of Digital Humanities, people had named it
>> Computational Humanities, like Computational Biology, for
>> example? (Yuk!)
> Be strong now: A short check on Google will confirm, that in Europe
> [even in KCL, by the way] the use of the terms "Computational
> Humanities", "Computational Literary Studies" etc. is rapidly expanding,
> gaining conferences, journals, chairs at universities and departments
> dedicated to it. If feel reassured by that developemtn, that I am not
> the only one believing, that the field I am identifying with has
> something to do with the usage of computers and the Humanities. The only
> reason, I am not advocating the term more enthusiastically is, that if
> you look at the program of most of the events labeled CH, you see a
> throwback into the nineties, when Humanities Computing had a tendency to
> concentrate almost exclusively on literary and linguistic stuff, which
> excluded so many new developments, that everybody was looking for a
> wider term, which covered other interesting things as well.
>
> Apologies to everybody, who is tired of this whole "what is DH"
> discussion. But I am afraid, we are all damned to repeat it until
> doomsday, until we replace a label having become meaningless by ... not
> another label!
>
> But possibly a discussion, what computational - or if you want: digital
> - technologies should do for the Humanities, how you can fairly evaluate
> achievements and how you can meaningfully teach it? And how as a
> consequence that one- / two-sentence question Willard asked for might
> look like? And what is a wonderful achievement and a goal worthy to
> dedicate your life to, but not part of "it"?
>
> Best regards,
> Manfred
_______________________________________________
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
