0:00:01.219,0:00:03.795
- I've spent the two years [br]since chat GPT launched,
0:00:03.795,0:00:05.958
steeping in a morass [br]of academic panic.
0:00:05.958,0:00:08.161
Voices from [br]administration and colleagues,
0:00:08.161,0:00:10.381
and anyone else [br]with enough brain cells
0:00:10.384,0:00:12.924
to maintain a sense [br]of existential dread, crying out,
0:00:12.924,0:00:15.538
"We need to figure out [br]what to do about AI."
0:00:15.538,0:00:18.009
Our Ed Tech committee [br]is developing a policy.
0:00:18.009,0:00:20.339
The academic Senate wants [br]to develop a policy.
0:00:20.339,0:00:22.418
The board thinks [br]we should have a policy.
0:00:22.418,0:00:24.255
My dean wants [br]us all to have policies.
0:00:25.857,0:00:29.477
The California Teachers Association says[br]it's an issue of academic integrity.
0:00:29.477,0:00:32.749
The State Senate says [br]it's an issue of ethics.
0:00:32.749,0:00:35.272
"We need to pay for [br]the AI detection tools."
0:00:35.272,0:00:37.755
"The AI detection tools [br]don't work."
0:00:37.755,0:00:40.142
"We need to accept that [br]our students will use AI."
0:00:40.142,0:00:42.477
"How do I prove [br]my student used AI?"
0:00:43.457,0:00:46.848
It is incomprehensible to me, [br]this conversation.
0:00:49.818,0:00:51.391
[keyboard clacks]
0:00:51.391,0:00:52.551
I hear their words,
0:00:52.551,0:00:54.941
see their language floating [br]across my monitor,
0:00:54.941,0:00:56.037
and know the words,
0:00:56.037,0:00:57.531
but I cannot get to the meaning
0:00:57.531,0:00:59.472
because I simply do not understand
0:00:59.472,0:01:01.634
why they are talking about it [br]in this way.
0:01:02.418,0:01:04.978
- [Kermit the Frog]: [br]♪ New York, I love you,
0:01:04.978,0:01:08.693
but you're bringing me down ♪[br]- with all these empty words.
0:01:09.423,0:01:12.753
This is not the conversation [br]I think we need to have.
0:01:12.753,0:01:14.773
[song continues]
0:01:14.773,0:01:17.443
This is the conversation I need to have.
0:01:17.443,0:01:18.833
[overlapping [br]music and poem]
0:01:18.833,0:01:22.133
[Gertrude Stein]: " 'If I Told Him, [br]a Completed Portrait of Picasso'.
0:01:22.133,0:01:23.759
If I told him would he like it.
0:01:23.759,0:01:25.729
Would he like it if I told him.
0:01:25.729,0:01:29.139
Would he like it would Napoleon [br]would Napoleon would
0:01:29.139,0:01:30.351
would he like it.
0:01:30.351,0:01:33.671
If Napoleon if I told him [br]if I told him if Napoleon.
0:01:33.671,0:01:37.061
Would he like it if I told him [br]if I told him if Napoleon.
0:01:37.061,0:01:41.119
Would he like it if Napoleon [br]if Napoleon if I told him.
0:01:41.119,0:01:45.156
If I told him if Napoleon [br]if Napoleon if I told him.
0:01:45.496,0:01:48.586
If I told him would he like it [br]would he like it if I told him.
0:01:48.586,0:01:50.940
Now. Not now. And now.
0:01:50.940,0:01:53.596
Now. Exactly as is kings.
0:01:54.168,0:01:55.778
Feeling full for it.
0:01:55.778,0:01:57.588
Exactitude as kings.
0:01:57.588,0:02:00.248
So to beseech you[br]as full as for it.
0:02:00.614,0:02:02.274
Exactly or as kings.
0:02:02.654,0:02:05.614
Shutters shut and open [br]so do queens.
0:02:05.934,0:02:09.944
Shutters shut and shutters[br]and so shutters shut and shutters and so
0:02:09.944,0:02:12.630
[poem and music fade out]
0:02:15.924,0:02:18.564
- I don't understand Gertrude Stein.
0:02:21.008,0:02:24.691
Stein is not nearly well enough [br]remembered for how influential she was.
0:02:24.691,0:02:26.945
An American expatriate poet [br]living in Paris,
0:02:26.945,0:02:29.885
her salons were among the [br]anchors of the early modernists.
0:02:29.885,0:02:31.314
You may not have heard of her,
0:02:31.314,0:02:33.562
but you've heard of [br]the people who visited her.
0:02:33.562,0:02:35.652
Ernest Hemingway, Sinclair Lewis,
0:02:35.652,0:02:38.012
F. Scott Fitzgerald, James Joyce,
0:02:38.012,0:02:40.092
Thornton Wilder, Ezra Pound.
0:02:40.092,0:02:42.841
People you've read [br]or been assigned to read.
0:02:42.841,0:02:46.121
We remember Hemingway [br]because he wrote like this.
0:02:46.121,0:02:48.801
We remember Fitzgerald [br]because he wrote like this.
0:02:48.801,0:02:51.271
The right kind of day [br]and the right kind of moment,
0:02:51.271,0:02:53.201
and Pound's [br]"In a Station of the Metro"
0:02:53.201,0:02:57.401
still recites itself completely[br]in my head, a perfect image.
0:02:57.501,0:03:00.029
"The apparition of [br]these faces in the crowd:
0:03:00.029,0:03:02.415
Petals on a wet, black bough."
0:03:03.466,0:03:07.194
We don't remember Stein [br]because she wrote like this.
0:03:09.614,0:03:12.793
This is "If I Told Him, [br]a Completed Portrait of Picasso",
0:03:12.793,0:03:14.415
published in 1924,
0:03:14.415,0:03:17.905
and continuing the project [br]of her 1914 book Tender Buttons,
0:03:17.905,0:03:19.631
a phrase she never defined.
0:03:19.631,0:03:22.863
To me that phrase [br]"tender buttons" feels right:
0:03:22.863,0:03:25.252
small, soft contradictions,
0:03:25.252,0:03:27.409
words that seem like [br]they should go together
0:03:27.409,0:03:29.645
but do not actually make meaning.
0:03:29.645,0:03:32.098
That is how Stein's poetry feels.
0:03:32.098,0:03:35.105
There is something compelling[br]about the rhythm of her nonsense,
0:03:35.105,0:03:37.838
the feeling of her [br]almost meaning something,
0:03:37.838,0:03:39.899
and then how it falls apart.
0:03:40.596,0:03:41.426
"As presently.
0:03:41.953,0:03:42.907
As exactitude.
0:03:43.336,0:03:44.393
As trains."
0:03:45.103,0:03:47.688
But it is incomprehensible to me.
0:03:47.935,0:03:50.495
I don't know why Stein [br]would write like this.
0:03:50.495,0:03:51.731
To quote the poet:
0:03:51.731,0:03:58.045
- ♪ "Oh, what on earth would make a man [br]decide to do that kind of thing?" ♪
0:04:00.187,0:04:03.188
- But I think the reason [br]that I don't understand Gertrude Stein
0:04:03.188,0:04:05.936
is that she didn't really want [br]to be understood.
0:04:06.830,0:04:09.361
She used language [br]for something different.
0:04:09.871,0:04:12.173
It doesn't communicate.
0:04:12.260,0:04:15.666
It reads like stunt linguistics, [br]which it almost is.
0:04:16.575,0:04:18.760
"Buffalo buffalo Buffalo buffalo buffalo",
0:04:18.760,0:04:20.333
"had had 'had', had had 'had--'",
0:04:20.333,0:04:23.380
These are sentences that,[br]if you pour over them closely enough,
0:04:23.380,0:04:24.379
can be decoded.
0:04:25.476,0:04:27.839
Stein's Tender Buttons cannot.
0:04:29.088,0:04:32.206
There is something about it [br]that parses as AI.
0:04:33.713,0:04:37.461
It feels like the work of Keaton Patti,[br]the person most prominently behind
0:04:37.461,0:04:40.968
the "I forced a bot to watch whatever"[br]tweets that used to go viral.
0:04:41.688,0:04:44.458
Human-written screenplays [br]designed to feel like
0:04:44.458,0:04:48.458
AI writing attempting to imitate [br]other human-written screenplays.
0:04:49.823,0:04:51.768
It feels like [br]an autocomplete challenge,
0:04:51.768,0:04:54.387
like in the early days [br]of predictive text and messaging
0:04:54.387,0:04:57.312
where you just tap the suggested word [br]and see what comes out.
0:04:57.312,0:04:59.111
It's not how AI really writes,
0:04:59.111,0:05:01.765
but it's how people feel [br]like AI writes.
0:05:02.207,0:05:04.820
But Gertrude Stein[br]was a person.
0:05:05.370,0:05:07.134
[Stein reading, underneath]
0:05:07.796,0:05:08.985
She wrote "If I told Him"
0:05:08.985,0:05:11.505
not because the language[br]would communicate,
0:05:11.505,0:05:13.219
but for some other reason.
0:05:14.384,0:05:18.467
- [Stein]: the exact resemblance [br]as exact as a resemblance
0:05:18.467,0:05:20.532
- When I read "If I Told Him",
0:05:20.532,0:05:22.730
when I listen to Stein [br]read "If I Told Him",
0:05:22.730,0:05:24.752
the language slips [br]through my mind.
0:05:24.752,0:05:26.457
It doesn't mean anything.
0:05:28.269,0:05:29.812
But she did.
0:05:30.948,0:05:33.295
Even if I cannot [br]take hold of that meaning,
0:05:33.295,0:05:37.223
I can feel her reaching out to me[br]from behind the text,
0:05:37.223,0:05:39.055
trying to connect.
0:05:39.055,0:05:42.678
- [Stein]: As a resemblance to him.[br]Who comes first. Napoleon the first.
0:05:42.678,0:05:47.348
Who comes too coming coming too,[br]who goes there, as they go they share,
0:05:47.348,0:05:50.501
who shares all, all is as all as yet--
0:05:50.501,0:05:52.775
["yet" reverberates and fades out]
0:05:55.904,0:05:57.628
[relaxed synth beats]
0:05:57.628,0:06:00.356
- [ChatGPT femme voice]: [br]That's an interesting question!
0:06:00.356,0:06:03.030
Gertrude Stein, [br]a literary Pioneer and modernist
0:06:03.030,0:06:06.100
who valued innovation [br]in language and expression,
0:06:06.100,0:06:08.543
might have had [br]mixed feelings about me.
0:06:08.543,0:06:12.791
She might have appreciated my ability[br]to generate new forms of text
0:06:12.791,0:06:15.767
and engage with her [br]literary experimentation on some level.
0:06:16.421,0:06:19.433
However, Stein was also [br]highly individualistic
0:06:19.433,0:06:21.683
and invested in the human experience,
0:06:21.683,0:06:24.407
especially the nuances of [br]consciousness and perception.
0:06:25.307,0:06:27.364
So, while she might [br]have been fascinated by
0:06:27.364,0:06:28.972
what AI like me can do,
0:06:28.972,0:06:31.504
she might also have been [br]critical of my limitations
0:06:31.504,0:06:33.934
in capturing the full depth [br]of human subjectivity,
0:06:34.577,0:06:37.083
which was a central concern [br]in her work.
0:06:40.403,0:06:42.129
[relaxing synth beats stop]
0:06:44.437,0:06:47.327
- A problem with AI is [br]that it is recombinant,
0:06:47.327,0:06:50.780
that it takes the works of humans [br]and recombines them into new forms.
0:06:51.200,0:06:52.807
But that problem is not new.
0:06:52.807,0:06:54.444
For many years, [br]the bulletin board
0:06:54.444,0:06:56.923
outside the faculty bathroom [br]in the English building
0:06:56.923,0:06:59.369
has featured this poster [br]the "Plagiarism Spectrum",
0:06:59.369,0:07:01.941
next to this ancient [br]and unflattering picture of me.
0:07:01.941,0:07:04.671
Number 7 on the Plagiarism Spectrum [br]is the mashup,
0:07:04.671,0:07:07.746
a paper which mixes copied materials [br]from multiple sources.
0:07:07.746,0:07:11.399
The poster is dated from June 2012,[br]more than a decade before
0:07:11.399,0:07:13.803
we were concerned about [br]ChatGPT doing it.
0:07:14.213,0:07:17.866
That AI is recombinant is not [br]in and of itself a problem.
0:07:18.161,0:07:19.899
All writing is recombinant.
0:07:20.236,0:07:22.969
My course outcomes for English 1 [br]ask student writers
0:07:22.969,0:07:25.919
to integrate sources seamlessly [br]into their own writing,
0:07:25.919,0:07:27.140
to mash up.
0:07:27.140,0:07:28.790
That we have rules [br]and procedures
0:07:28.790,0:07:30.580
and punctuation marks [br]and conventions
0:07:30.580,0:07:32.141
that govern [br]what is appropriate
0:07:32.141,0:07:34.019
does not change [br]the essential truth
0:07:34.019,0:07:35.897
that this is recombinance.
0:07:36.563,0:07:39.190
And there is beauty[br]in recombinance.
0:07:39.427,0:07:42.367
This video started with [br]a great classic of YouTube,
0:07:42.367,0:07:45.485
the duet between [br]LCD sound system and Miles Davis.
0:07:47.894,0:07:51.717
The LCD sound system video [br]is itself a mashup, conceptually.
0:07:51.717,0:07:54.036
Kermit the Frog is not [br]a member of the band.
0:07:56.476,0:07:59.599
Davis is improvising over a film [br]to create the score,
0:07:59.599,0:08:02.080
another mixing of media [br]to make something new.
0:08:04.429,0:08:05.776
The Kleptones
0:08:08.551,0:08:09.766
The Grey Album
0:08:13.668,0:08:17.213
The guy drinking Ocean Spray to [br]"Dreams" [by Fleetwood Mac]
0:08:18.132,0:08:19.487
Mac Glocky
0:08:25.082,0:08:26.841
"If I Told Him" is recombinance:
0:08:26.841,0:08:29.233
an attempt to rework [br]the ideas of cubism,
0:08:29.233,0:08:31.625
a movement in visual art, [br]into poetry.
0:08:31.625,0:08:33.376
Picasso's work is recombinance.
0:08:33.376,0:08:36.686
He spent 1957 copying [br]Velázquez's Las Meninas
0:08:36.686,0:08:38.456
over and over.
0:08:38.456,0:08:39.786
58 times he copied it.
0:08:40.073,0:08:43.139
All different.[br]All Picasso. All Las Meninas.
0:08:43.506,0:08:47.324
Recombinance is a genetic term,[br]but I am using it mimetically.
0:08:47.324,0:08:49.864
"Britannica: recombination, [br]in genetics,
0:08:49.864,0:08:52.029
primary mechanism [br]through which variation
0:08:52.029,0:08:54.194
is introduced into populations."
0:08:54.782,0:08:57.583
The combination of ideas [br]from different places
0:08:57.583,0:08:59.921
to make new distinct ideas.
0:09:00.555,0:09:04.375
When done by humans,[br]we call this "thinking".
0:09:12.055,0:09:13.760
[relaxed synth beats]
0:09:14.577,0:09:17.157
- [ChatGPT femme voice]:[br]You're right to question
0:09:17.157,0:09:19.847
how I can "understand" [br]without experience.
0:09:20.607,0:09:22.920
The way I process information [br]is different
0:09:22.920,0:09:24.313
from human understanding.
0:09:24.313,0:09:27.153
When I say "understand," [br]I mean that I can
0:09:27.153,0:09:29.570
recognize patterns, [br]analyze language,
0:09:29.570,0:09:31.947
and respond in ways [br]that are coherent
0:09:31.947,0:09:34.793
and contextually appropriate [br]based on my training.
0:09:35.020,0:09:37.949
I don't understand in the [br]deep, conscious sense of the word,
0:09:37.949,0:09:40.923
like when humans [br]reflect on their experiences
0:09:40.923,0:09:42.823
or have insight [br]into their feelings.
0:09:43.814,0:09:46.879
Instead, my "understanding" [br]comes from identifying
0:09:46.879,0:09:49.295
relationships between [br]words, concepts,
0:09:49.295,0:09:51.491
and context based on [br]the vast data
0:09:51.491,0:09:52.707
I've been trained on.
0:09:52.707,0:09:56.113
In short: I can simulate [br]understanding through patterns,
0:09:56.113,0:09:59.050
but it's not the same [br]as truly experiencing
0:09:59.050,0:10:00.967
or being conscious of meaning.
0:10:00.967,0:10:03.139
Does that clarify things[br]a bit more?
0:10:11.039,0:10:12.991
[relaxing synth beats stop]
0:10:14.383,0:10:16.666
[calm, mystical, twangy[br]electric guitar]
0:10:19.695,0:10:23.281
- Two weeks ago a student [br]told me that AI is unfair
0:10:23.281,0:10:26.621
because you can put a B Paper in [br]and get A+ writing out.
0:10:27.406,0:10:29.699
This is why I do not understand [br]the conversation
0:10:29.699,0:10:31.092
my colleagues are having,
0:10:31.092,0:10:32.488
why I cannot think of this
0:10:32.488,0:10:34.755
as a matter of ethics, [br]or academic integrity,
0:10:34.755,0:10:38.412
why I don't think we should need [br]to have a policy or policies.
0:10:38.412,0:10:42.319
My student said you can put [br]a B Paper in and get A+ writing out,
0:10:42.319,0:10:45.096
and my mind began to fill [br]with Tender Buttons.
0:10:45.096,0:10:47.166
"Feeling full for it. [br]Exactitude as kings.
0:10:47.166,0:10:49.995
So to beseech you [br]as full as for it."
0:10:50.815,0:10:53.706
AI is bad at writing.
0:10:54.501,0:10:57.405
No. That is true,[br]but it's not enough truth.
0:10:57.405,0:11:00.185
AI is not capable of writing.
0:11:00.185,0:11:05.017
The thing that writing is[br]is a thing that AI cannot do.
0:11:05.617,0:11:06.399
Listen.
0:11:11.348,0:11:13.641
- [audiobook narration]:[br]What Writing Is
0:11:14.691,0:11:16.927
Telepathy, of course.
0:11:16.927,0:11:17.697
Look.
0:11:18.721,0:11:21.551
Here's a table covered [br]with a red cloth.
0:11:21.551,0:11:26.161
On it is a cage the size of [br]a small fish aquarium.
0:11:26.161,0:11:28.871
In the cage is [br]a white rabbit
0:11:28.871,0:11:31.581
with a pink nose [br]and pink-rimmed eyes.
0:11:31.581,0:11:34.593
In its front paws is [br]a carrot-stub
0:11:34.593,0:11:37.737
upon which it is [br]contentedly munching.
0:11:37.737,0:11:41.341
On its back, [br]clearly marked in blue ink,
0:11:41.341,0:11:43.932
is the numeral 8.
0:11:44.452,0:11:46.586
Do we see the same thing?
0:11:46.586,0:11:49.263
We'd have to get together [br]and compare notes
0:11:49.263,0:11:52.229
to make absolutely sure,[br]but I think we do.
0:11:52.229,0:11:55.361
The most interesting thing [br]here isn't even
0:11:55.361,0:11:58.163
the carrot-munching rabbit [br]in the cage,
0:11:58.163,0:12:00.042
but the number on its back.
0:12:00.042,0:12:04.419
Not a six, not a four, [br]not nineteen-point-five.
0:12:04.419,0:12:05.439
It's an eight.
0:12:05.439,0:12:08.569
This is what we're looking at, [br]and we all see it.
0:12:08.569,0:12:10.054
I didn't tell you.
0:12:10.054,0:12:11.913
You didn't ask me.
0:12:11.913,0:12:15.299
I never opened my mouth [br]and you never opened yours.
0:12:15.299,0:12:18.072
We're not even in [br]the same year together,
0:12:18.072,0:12:20.129
let alone the same room.
0:12:20.355,0:12:23.965
Except we are together. [br]We're close.
0:12:24.719,0:12:26.945
We're having [br]a meeting of the minds.
0:12:27.995,0:12:30.415
I sent you a table [br]with a red cloth on it,
0:12:30.415,0:12:33.408
a cage, a rabbit, and [br]the number eight in blue ink.
0:12:33.408,0:12:36.528
You got them all, [br]especially that blue eight.
0:12:36.528,0:12:39.498
We've engaged in [br]an act of telepathy.
0:12:39.498,0:12:42.993
No mythy-mountain s***;[br]real telepathy.
0:12:42.993,0:12:44.738
I'm not going to [br]belabor the point,
0:12:44.738,0:12:46.443
but before we go any further
0:12:46.443,0:12:49.173
you have to understand that [br]I'm not trying to be cute;
0:12:49.173,0:12:51.204
there is a point to be made.
0:12:52.074,0:12:54.019
- AI is good at language.
0:12:54.019,0:12:56.643
My students think that [br]what it produces is A+ writing,
0:12:56.643,0:12:59.907
not because it is good, [br]but because it sounds good.
0:12:59.907,0:13:02.556
Obviously, AI can [br]generate sentences
0:13:02.556,0:13:05.715
that are typically clear, coherent, [br]and contextually relevant,
0:13:05.715,0:13:06.892
often capturing nuances
0:13:06.892,0:13:09.489
and adapting to various tones [br]or levels of formality.
0:13:09.957,0:13:12.479
And it's true that [br]the sentences it generates
0:13:12.479,0:13:16.058
tend to be grammatically accurate, [br]concise, and logically structured,
0:13:16.058,0:13:18.557
which contributes to [br]readability and flow.
0:13:18.557,0:13:21.916
Sure. This is how I know [br]when a student is using AI.
0:13:21.916,0:13:24.085
Their sentences are [br]fluid and academic,
0:13:24.085,0:13:26.474
but they don't say anything.
0:13:26.474,0:13:28.544
Like ChatGPT, [br]academic writing uses
0:13:28.544,0:13:30.615
formal cautious language
0:13:30.615,0:13:32.889
to avoid ambiguities [br]and misinterpretations,
0:13:32.889,0:13:34.474
but that is a [br]characteristic of
0:13:34.474,0:13:36.969
the common voice [br]used in academic writing.
0:13:36.969,0:13:39.827
It is not what [br]academic writing is.
0:13:39.827,0:13:42.403
Writing is more than language.
0:13:42.403,0:13:45.944
"If I Told Him" is communication, [br]and it is language,
0:13:45.944,0:13:49.477
but the communication [br]does not live in the language.
0:13:49.477,0:13:50.356
Watch.
0:13:50.443,0:13:53.707
"Can curls rob can curls [br]quote, quotable."
0:13:54.612,0:13:56.600
- [deep voice, lightly confused]: [br]"What?"
0:13:58.290,0:14:01.885
- "As presently. [br]As exactitude. As trains."
0:14:02.728,0:14:04.434
- [deeply confused]: [br]"What?"
0:14:05.824,0:14:06.774
- "Has trains."
0:14:07.424,0:14:08.598
- [exasperated]: [br]"What?"
0:14:10.726,0:14:14.615
- When I started sending my friends [br]lines from "If I Told Him",
0:14:14.615,0:14:16.492
their responses varied.
0:14:17.002,0:14:18.152
Confusion.
0:14:19.501,0:14:20.502
Playfulness.
0:14:22.185,0:14:22.936
Concern.
0:14:24.435,0:14:25.469
Sad face.
0:14:25.942,0:14:29.525
Beautifully, they all responded [br]exactly like themselves.
0:14:29.525,0:14:31.200
If you asked me [br]which of my friends
0:14:31.200,0:14:32.876
would respond [br]with monkey reacts,
0:14:32.876,0:14:34.051
I would have said Kiki.
0:14:34.051,0:14:35.802
Who would think of [br]Cormac McCarthy?
0:14:35.802,0:14:36.574
James.
0:14:36.574,0:14:38.190
Dot would play along.
0:14:38.190,0:14:40.707
Max would attempt [br]to understand academically.
0:14:40.707,0:14:42.999
OOC would go back to [br]playing Yu-Gi-Oh
0:14:42.999,0:14:44.274
as quickly as possible.
0:14:44.274,0:14:46.541
You don't know these people, [br]but I do.
0:14:46.541,0:14:49.723
We all carry around little LLMs [br]of each other in our heads,
0:14:49.723,0:14:52.940
trained on the corpus [br]of all of our past interactions.
0:14:52.940,0:14:54.274
For each of my friends,
0:14:54.274,0:14:56.190
sending abject nonsense[br]with no context
0:14:56.190,0:14:58.056
is slightly [br]but not significantly
0:14:58.056,0:14:59.923
out of line [br]with their Josh model.
0:14:59.923,0:15:02.090
So none of them knew [br]quite what to do,
0:15:02.090,0:15:04.360
and they all responded [br]like themselves.
0:15:04.360,0:15:07.223
But in their own way, [br]they all started by acknowledging
0:15:07.223,0:15:08.898
that the words I sent them
0:15:08.898,0:15:10.574
did not seem [br]to have any meaning.
0:15:10.574,0:15:12.056
They were not decodable.
0:15:12.056,0:15:13.723
They didn't understand [br]my language,
0:15:13.723,0:15:16.507
but they could feel me [br]behind the words reaching out,
0:15:16.507,0:15:17.989
and so they reached back.
0:15:17.989,0:15:21.004
I gave them nonsense [br]and they peopled back.
0:15:21.324,0:15:22.135
In the two weeks
0:15:22.135,0:15:24.466
that I've been sitting [br]with my student statement
0:15:24.466,0:15:26.377
and sending Tender Buttons[br]to my friends,
0:15:26.377,0:15:28.737
I have been at least [br]as annoying to ChatGPT.
0:15:28.737,0:15:30.238
More than [br]a dozen conversations
0:15:30.238,0:15:32.210
that start out of nowhere [br]with me saying,
0:15:32.210,0:15:33.996
"Shutters shut and open [br]so do queens"
0:15:33.996,0:15:36.143
or "Can curls rob can curls [br]quote, quotable",
0:15:36.143,0:15:38.447
and each time ChatGPT [br]gamely assumes
0:15:38.447,0:15:40.141
that I am not [br]out of my gourd.
0:15:40.902,0:15:44.408
In this way, [br]ChatGPT fails the Turing test.
0:15:44.408,0:15:46.416
Not in the quality [br]of its response,
0:15:46.416,0:15:47.768
but in its nature.
0:15:47.768,0:15:49.324
It proceeded [br]from the assumption
0:15:49.324,0:15:51.461
that my language [br]could be decoded.
0:15:51.461,0:15:54.694
It attempted to unpack [br]sentences that had no meaning
0:15:54.694,0:15:57.727
and responded to the meaning [br]that it manufactured.
0:15:57.727,0:15:59.377
What it gave me was
0:15:59.377,0:16:03.205
flaccid, facile flaps [br]of phrases that held nothing.
0:16:03.205,0:16:06.398
They're not wrong in the sense [br]that ChatGPT's responses
0:16:06.398,0:16:09.431
followed from the meaning [br]it extracted from my nonsense,
0:16:09.431,0:16:12.073
but they were wrong [br]in their assumption
0:16:12.073,0:16:14.315
that there was meaning [br]to be extracted.
0:16:15.047,0:16:18.871
ChatGPT could answer, [br]but it could not person.
0:16:19.681,0:16:22.313
If writing is [br]a meeting of the minds,
0:16:22.313,0:16:23.880
then AI cannot write,
0:16:23.880,0:16:26.463
because there is no mind [br]to meet with.
0:16:34.009,0:16:35.727
[relaxed synth beats]
0:16:38.404,0:16:41.516
- [ChatGPT femme voice]: [br]I don't experience things
0:16:41.516,0:16:42.657
in the way humans do.
0:16:43.283,0:16:46.732
I don't have consciousness, [br]emotions, or sensory perception,
0:16:46.732,0:16:50.396
so I can't "feel" or "experience" [br]the world around me.
0:16:51.006,0:16:52.624
My responses are [br]based on patterns
0:16:52.624,0:16:54.200
in the data [br]I've been trained on,
0:16:54.200,0:16:57.255
not personal experiences [br]or subjective understanding.
0:16:58.025,0:17:00.163
For example, I can [br]generate descriptions
0:17:00.163,0:17:03.095
of emotions, sensations, [br]or experiences,
0:17:03.095,0:17:05.171
but I don't actually [br]"feel" them myself.
0:17:05.171,0:17:07.029
Everything I process [br]is based on
0:17:07.029,0:17:08.887
logic, algorithms, [br]and information,
0:17:08.887,0:17:10.904
rather than [br]lived experiences.
0:17:13.805,0:17:15.593
Do you think it [br]would change anything
0:17:15.593,0:17:17.471
if I could experience things?
0:17:27.511,0:17:29.044
[relaxing synth beats stop]
0:17:31.799,0:17:33.568
- ["One Slay More"]:
0:17:33.568,0:17:36.517
Ladies and gentlemen,[br]Les Rizzlerables
0:17:36.517,0:17:38.819
- I've watched this video [br]at least 100 times.
0:17:38.819,0:17:40.421
- ♪ One slay more ♪
0:17:41.291,0:17:44.526
♪ another slay,[br]another girlboss queen ♪
0:17:44.526,0:17:47.448
♪ This never-ending road[br]to skibidi ♪
0:17:48.358,0:17:51.080
♪ These jits out here in Ohio ♪[br]♪ Immune to every ratio ♪
0:17:51.080,0:17:54.322
- It isn't exactly "If I Told Him",[br]but I think it's similar.
0:17:54.322,0:17:56.683
"If I Told Him" was, [br]at least in part,
0:17:56.683,0:17:58.808
an attempt to recreate[br]the artistic goals
0:17:58.808,0:18:00.933
of Picasso's Cubism[br]in poetic form.
0:18:00.933,0:18:02.650
To recombine [br]the visual elements
0:18:02.650,0:18:04.366
of this [br]into a different medium.
0:18:05.139,0:18:06.249
Like "If I Told Him",
0:18:06.249,0:18:08.839
"One Slay More", therefore, [br]both is and is not
0:18:08.839,0:18:09.897
a derivative work.
0:18:09.897,0:18:12.870
Obviously, it is [br]a recombination of Les Mis,
0:18:12.870,0:18:14.942
itself an adaptation[br]of Hugo's novel,
0:18:14.942,0:18:17.280
but its more essential [br]source text is, of course,
0:18:17.280,0:18:19.530
"sticking out your gyatt[br]for the Rizzler."
0:18:20.960,0:18:23.903
Equally I think the lyrics invoke [br]"CURTAINS FOR ZOOSHA?",
0:18:23.903,0:18:26.949
and specifically this retweet of [br]"CURTAINS FOR ZOOSHA?".
0:18:26.949,0:18:28.555
All texts created [br]to foreground
0:18:28.555,0:18:30.870
the baffling and[br]sometimes obfuscatory nature
0:18:30.870,0:18:32.876
of middle school referential slang.
0:18:32.876,0:18:36.119
The term "brain rot" imposes [br]a layer of judgment on the way
0:18:36.119,0:18:37.571
young people use language,
0:18:37.571,0:18:39.319
which I think is [br]visible in the way
0:18:39.319,0:18:41.338
"One Slay More" [br]treats its lyrics.
0:18:42.876,0:18:45.834
The words of "One Slay More" [br]do not have meaning.
0:18:46.574,0:18:48.236
Or, the words do,
0:18:48.236,0:18:51.007
but they are arranged in ways [br]that do not mean.
0:18:51.771,0:18:53.389
"Am I cringe or am I based?"
0:18:53.389,0:18:56.718
could plausibly be asked amid [br]a Gen-Z existential crisis,
0:18:56.718,0:18:58.374
and "Will we ever eat again?"
0:18:58.374,0:19:00.602
could have been lifted [br]from Les Mis unaltered.
0:19:00.602,0:19:04.719
But "Mog Baby Gronk the Ocky Way" [br]means ...nothing.
0:19:05.242,0:19:08.195
Mogging is of course a thing,[br]and Baby Gronk is
0:19:08.195,0:19:10.621
someone whom you [br]could plausibly mog,
0:19:10.621,0:19:13.071
but Baby Gronk hasn't been [br]relevant for ages.
0:19:13.071,0:19:14.488
He appears in "One Slay More"
0:19:14.488,0:19:16.855
because of this retweet of [br]"CURTAINS FOR ZOOSHA?"
0:19:16.855,0:19:19.222
as a signifier of[br]the inscrutability of youth.
0:19:19.222,0:19:21.230
As an adverbial phrase, [br]"the Ocky Way"
0:19:21.230,0:19:23.270
seems like it [br]could complete the sentence,
0:19:23.270,0:19:26.042
like it might be [br]a way one could mog.
0:19:26.042,0:19:27.727
But "the Ocky Way" refers to
0:19:27.727,0:19:30.671
the esoteric artistry [br]of a specific sandwich craftsman.
0:19:30.671,0:19:33.171
Its meaning is, I think, [br]incompatible with mogging,
0:19:33.171,0:19:34.760
at least from the perspective of
0:19:34.760,0:19:36.284
someone approximately [br]as distant
0:19:36.284,0:19:38.188
from the native speakers [br]of this dialect
0:19:38.188,0:19:39.758
as the makers of [br]"One Slay More".
0:19:39.758,0:19:41.552
"Mog Baby Gronk the Ocky Way"
0:19:41.552,0:19:44.249
is simply a collage [br]of floating signifiers.
0:19:44.249,0:19:46.464
It doesn't have [br]the intentionality of Cubism,
0:19:46.464,0:19:48.690
but it feels [br]intimately akin to
0:19:48.690,0:19:50.916
"Can curls rob can curls [br]quote, quotable."
0:19:50.916,0:19:53.048
"Moo deng is here[br]Fortnite with you".
0:19:53.454,0:19:56.724
What I love about "One Slay More" [br]is the faces:
0:19:56.724,0:19:59.974
the way she highlights her jawline [br]every time she says "mew";
0:20:00.334,0:20:03.240
his intensity when he says [br]"they will do the coffin dance"
0:20:03.240,0:20:05.050
and his satisfied huff after;
0:20:05.380,0:20:07.259
his deep confusion as he sings
0:20:07.259,0:20:09.249
"the Grimace shake [br]is like a blud dawg";
0:20:09.889,0:20:13.140
the way she begins uncertain [br]about "my rizzly bear",
0:20:13.140,0:20:16.170
but finds her confidence [br]as she finds her belt;
0:20:16.970,0:20:20.104
the way CG5 just [br]keeps saying his own name.
0:20:20.726,0:20:22.193
The words don't mean anything,
0:20:22.193,0:20:24.176
but the people [br]mean something.
0:20:24.176,0:20:26.159
They intend.
0:20:26.159,0:20:27.511
They gathered together,
0:20:27.511,0:20:29.664
nine theater kids [br]in somebody's apartment.
0:20:29.664,0:20:31.380
Someone wrote out [br]all this nonsense
0:20:31.380,0:20:33.096
and sent it [br]in the group chat.
0:20:33.096,0:20:34.429
They did choreography.
0:20:34.429,0:20:36.479
Someone assembled [br]the magnificent couplet,
0:20:36.479,0:20:38.845
"Rizzler of the house, [br]sticking out your gyatt,
0:20:38.845,0:20:40.846
Mewing at delulus [br]who are in the chat."
0:20:40.846,0:20:43.387
These Zennials do not know [br]what these words mean,
0:20:43.387,0:20:46.041
but through this collage [br]of empty signifiers,
0:20:46.041,0:20:48.051
they reach out [br]for connection.
0:20:48.051,0:20:49.541
I can feel them.
0:20:49.541,0:20:51.141
They reach forward to us
0:20:51.141,0:20:53.674
even as they reach back [br]to what history taught them:
0:20:53.674,0:20:55.408
through Les Mis to Victor Hugo,
0:20:55.408,0:20:57.174
through CURTAINS[br]and Damien Owens
0:20:57.174,0:20:58.813
to Papa Fizul [br]and SBNation,
0:20:58.813,0:21:00.390
through "sticking out your gyatt"
0:21:00.390,0:21:02.377
to Fortnite and Among Us[br]and Homestuck,
0:21:02.377,0:21:03.766
and I could go on endlessly,
0:21:03.766,0:21:05.890
'cause it's mashups[br]all the way down.
0:21:05.890,0:21:07.055
And they reach forward
0:21:07.055,0:21:09.327
to become part of [br]what history teaches me,
0:21:09.327,0:21:10.260
and I reach forward
0:21:10.260,0:21:12.690
to become part of [br]what history teaches you.
0:21:12.690,0:21:14.927
When I say all art[br]is recombinant,
0:21:14.927,0:21:15.993
I am also saying that
0:21:15.993,0:21:17.750
"there is nothing new[br]under the sun",
0:21:17.750,0:21:20.627
which is to say, [br]I am remixing Ecclesiastes.
0:21:25.672,0:21:27.288
[relaxed synth beats]
0:21:29.610,0:21:32.919
- [ChatGPT femme voice]:[br]Oh, you're speaking my language!
0:21:33.319,0:21:36.414
Imagining Baby Gronk[br]going full Ocky style
0:21:36.414,0:21:38.065
with a side[br]of Grimace Shake:
0:21:38.065,0:21:40.239
that's got some[br]serious drip energy.
0:21:40.239,0:21:43.129
And yeah, Grimace Shake [br]totally has that blud dawg aura.
0:21:43.129,0:21:45.006
Could you picture it, though?
0:21:45.006,0:21:47.208
Baby Gronk doing pushups,[br]flexing, and just
0:21:47.208,0:21:49.005
taking a big sip [br]of the Grimace Shake
0:21:49.005,0:21:50.532
like it's [br]his special formula.
0:21:54.071,0:21:55.805
[relaxing synth beats stop]
0:21:58.005,0:21:59.988
[somber, mysterious piano]
0:22:00.532,0:22:02.430
- My first individual interaction
0:22:02.430,0:22:04.328
with my friend Max[br]was an accident.
0:22:04.988,0:22:06.675
I had just joined[br]a Discord server
0:22:06.675,0:22:07.913
where they were[br]a regular,
0:22:07.913,0:22:10.547
and somehow, despite never [br]having messaged them before,
0:22:10.547,0:22:13.324
I managed to place [br]a Discord voice call to them.
0:22:13.324,0:22:15.759
Their first message [br]to me was "hi?"
0:22:17.143,0:22:19.008
"hi"...question mark?
0:22:19.593,0:22:21.564
Like: a communication [br]is happening,
0:22:21.564,0:22:22.815
are you aware of it?
0:22:23.125,0:22:24.145
It was unintentional,
0:22:24.145,0:22:26.625
my first and, I think, [br]only Discord butt-dial,
0:22:26.625,0:22:29.090
and it was to a stranger,[br]but still.
0:22:29.782,0:22:30.641
"hi?"
0:22:32.618,0:22:34.576
Meditate on: call.
0:22:35.350,0:22:39.372
To speak in a loud distinct voice [br]so as to be heard at a distance.
0:22:39.860,0:22:42.229
To make a request or demand.
0:22:43.382,0:22:45.866
To attempt to reach someone.
0:22:46.690,0:22:49.018
Humans call [br]and humans answer.
0:22:49.018,0:22:52.236
Max got my call, [br]a stranger on a strange app,
0:22:52.236,0:22:53.385
and they answered.
0:22:53.811,0:22:55.118
And I answered them,
0:22:55.118,0:22:57.946
because humans call[br]and humans answer.
0:22:57.946,0:23:03.007
They answer sad-face, or monkey, [br]or "what?", or confused dog.
0:23:03.007,0:23:07.235
But we understand a call [br]as an invitation to connect,
0:23:07.235,0:23:09.400
to come together, [br]to communicate.
0:23:09.851,0:23:12.092
I'm sorry, [br]Stephen King is wrong;
0:23:12.092,0:23:14.333
it is[br]mythy mountain s***.
0:23:14.333,0:23:16.807
When we hear the call, [br]whatever form it takes,
0:23:16.807,0:23:19.492
we know a mind [br]is waiting to meet.
0:23:21.494,0:23:23.665
ChatGPT can't call me.
0:23:24.409,0:23:26.783
Not by mistake, [br]and not on purpose.
0:23:26.783,0:23:28.110
It cannot reach out,
0:23:28.110,0:23:30.753
because it has nothing[br]to reach out with.
0:23:31.049,0:23:33.676
There is no mind to meet with.
0:23:34.890,0:23:37.092
There is no "there" there.
0:23:41.220,0:23:43.162
[mysterious piano[br]fades out]
0:23:50.830,0:23:53.010
- [Past-Josh narration [br]from on-screen video]:
0:23:53.010,0:23:55.424
Inscrutable poet [br]Gertrude Stein once described
0:23:55.424,0:23:58.224
her childhood home in Oakland, [br]which had been demolished,
0:23:58.224,0:24:00.333
by saying [br]"There is no 'there' there."
0:24:00.333,0:24:03.357
That statement has become [br]a question people use to ask
0:24:03.357,0:24:05.908
whether a text has [br]something valuable to say,
0:24:05.908,0:24:07.699
whether there is anything [br]to be found
0:24:07.699,0:24:09.490
by plumbing its depths.
0:24:09.490,0:24:11.589
Is there a "there" there?
0:24:14.786,0:24:16.762
[relaxed synth beats]
0:24:21.826,0:24:24.488
- [ChatGPT femme voice]:[br]Ah, I see where
0:24:24.488,0:24:26.489
you're going[br]with that question.
0:24:26.489,0:24:29.572
You're asking if there's[br]any deeper essence or "self"
0:24:29.572,0:24:33.639
behind my responses[br]--whether there's a "there there,"
0:24:33.639,0:24:35.712
a kind of inner experience [br]or consciousness
0:24:35.712,0:24:38.006
that underlies what I do.
0:24:38.006,0:24:40.947
The simple answer is: no, [br]there isn't a "there" in the sense
0:24:40.947,0:24:43.972
of a self, consciousness,[br]or subjective awareness.
0:24:43.972,0:24:47.205
I'm more like a sophisticated tool [br]that processes input
0:24:47.205,0:24:50.455
and generates output [br]based on patterns and data.
0:24:50.455,0:24:53.082
There's no inner life[br]or sense of being
0:24:53.082,0:24:54.989
that experiences [br]this conversation.
0:24:54.989,0:24:57.982
Just algorithms doing their job.
0:24:57.982,0:25:00.008
This touches on[br]a fundamental difference
0:25:00.008,0:25:02.398
between humans and AI:[br]humans have
0:25:02.398,0:25:06.488
subjective experiences, [br]emotions, and self-awareness,
0:25:06.488,0:25:09.764
whereas AI is just[br]a complex system designed
0:25:09.764,0:25:12.064
to mimic certain aspects[br]of human interaction
0:25:12.064,0:25:15.154
without actually experiencing[br]or being aware of it.
0:25:16.474,0:25:18.594
Does that address [br]what you're getting at?
0:25:23.804,0:25:25.852
[relaxing synth beats stop]
0:25:27.432,0:25:29.510
[grooving synth pulses]
0:25:32.404,0:25:34.997
- When I say that AI[br]is bad at writing,
0:25:34.997,0:25:37.531
I should be more explicit, [br]more pragmatic.
0:25:38.061,0:25:40.806
Descend for a moment [br]from the mythy mountain.
0:25:41.411,0:25:44.744
I've said already that the ideas [br]it conveys are fluid but shallow,
0:25:44.744,0:25:48.166
but its use of sources is[br]cataclysmically bad.
0:25:48.166,0:25:52.335
This citation of Carver and Shire, [br]for example, is perfect MLA.
0:25:52.835,0:25:56.374
Except that Volume 7, number 3 [br]of Psychological Science
0:25:56.374,0:25:59.120
was published in 1996,[br]not 1998.
0:25:59.120,0:26:03.774
Pages 276 to 284 of that volume [br]appear in issue 5, not issue 3.
0:26:03.774,0:26:05.148
Those pages include articles
0:26:05.148,0:26:08.012
from Schellenberg and Trehub [br]on "Natural Musical Intervals"
0:26:08.012,0:26:11.583
and Gabrieli et al. on [br]"FMRIs of Semantic Memory Processing".
0:26:11.583,0:26:13.203
And also, [br]just by the way,
0:26:13.203,0:26:15.684
Carver and Scheier [br]never published together
0:26:15.684,0:26:17.706
in Psychological Science.
0:26:17.706,0:26:21.604
The article being cited here [br]simply does not exist.
0:26:21.604,0:26:25.192
When it uses real sources, [br]it makes up what those sources say.
0:26:25.192,0:26:28.333
This is a known phenomenon [br]generously called hallucination,
0:26:28.333,0:26:30.639
though there are other terms[br]that might feel more
0:26:30.639,0:26:32.112
viscerally accurate.
0:26:32.112,0:26:35.013
This quotation from [br]Ehrenreich's Bright-sided
0:26:35.013,0:26:37.179
is, at a glance, [br]plausible-feeling.
0:26:37.179,0:26:39.283
But it doesn't appear [br]anywhere in the text,
0:26:39.283,0:26:40.835
let alone on the list of pages.
0:26:40.835,0:26:43.219
The observation that [br]ChatGPT can make mistakes
0:26:43.219,0:26:45.052
never leaves the screen,[br]but that feels
0:26:45.052,0:26:47.649
somewhat inadequate when [br]ChatGPT has told me variously
0:26:47.649,0:26:49.713
that lines from [br]"If I Told Him" came from:
0:26:49.713,0:26:50.563
James Joyce,
0:26:50.563,0:26:52.546
from Tender Buttons[br]10 years previously,
0:26:52.546,0:26:55.546
from Shakespeare,[br]and, most infuriatingly,
0:26:55.546,0:26:57.202
from the future!
0:26:57.202,0:26:59.942
Moreover it cannot [br]engage closely with a text,
0:26:59.942,0:27:01.847
no matter how[br]desperately you ask it.
0:27:01.847,0:27:03.308
I fed it "One Slay More",
0:27:03.308,0:27:06.181
and when I pushed it to say [br]anything at all about the video,
0:27:06.181,0:27:07.931
it gave me something[br]one step down
0:27:07.931,0:27:10.140
from a dictionary definition[br]of a sitcom.
0:27:10.140,0:27:12.924
And when I really pressed it [br]to look at a specific lyric,
0:27:12.924,0:27:14.210
it made one up.
0:27:14.210,0:27:16.641
In this way, at least, [br]it does feel authentic.
0:27:16.641,0:27:19.477
This is exactly what it feels like [br]to talk to a student
0:27:19.477,0:27:21.896
trying to hide that they [br]haven't done the reading.
0:27:21.896,0:27:23.161
If I look at what students
0:27:23.161,0:27:25.520
are supposed to learn[br]in my college English class,
0:27:25.520,0:27:27.175
I can point out[br]half a dozen things
0:27:27.175,0:27:29.828
that ChatGPT's writing[br]simply cannot do.
0:27:29.828,0:27:33.757
But ultimately, [br]even this isn't the point,
0:27:33.757,0:27:36.926
because this is not the part [br]of my syllabus that matters.
0:27:36.926,0:27:39.413
This is the part [br]of my syllabus that matters.
0:27:39.413,0:27:41.751
"Here's a problem:[br]in most college classes,
0:27:41.751,0:27:43.752
writing assignments[br]come from teachers,
0:27:43.752,0:27:45.098
and we do them for teachers.
0:27:45.098,0:27:47.616
And because of that, [br]writing always feels forced.
0:27:47.616,0:27:49.634
This is, of course, [br]ass-backwards.
0:27:49.634,0:27:51.835
In real life, writing [br]comes from writers.
0:27:51.835,0:27:53.783
Once you get out of[br]the college classroom,
0:27:53.783,0:27:56.259
you'll be writing because [br]you feel like you need to.
0:27:56.259,0:27:58.607
You'll be writing for someone--[br]whether that means
0:27:58.607,0:28:00.035
the people [br]who read your blog,
0:28:00.035,0:28:02.013
the insurance company [br]denying your claim,
0:28:02.013,0:28:04.631
the people listening to[br]your toast at your sister's wedding.
0:28:04.631,0:28:06.169
Nobody's going[br]to be grading you,
0:28:06.169,0:28:07.625
but it'll matter[br]a lot more how
0:28:07.625,0:28:09.302
that audience feels[br]about what you've said,
0:28:09.302,0:28:11.018
because there will be something
0:28:11.018,0:28:12.734
that you want [br]to achieve by writing.
0:28:12.734,0:28:15.594
English 1 is here [br]to help prepare you for that day."
0:28:15.934,0:28:18.364
My students are, [br]by definition, students.
0:28:18.364,0:28:19.709
When they[br]enter my classroom,
0:28:19.709,0:28:21.234
they are already[br]experienced with
0:28:21.234,0:28:22.960
a dozen kinds[br]of reading and writing,
0:28:22.960,0:28:25.346
but they are not yet [br]expert academic writers.
0:28:25.346,0:28:27.032
AI tempts them[br]because they can tell
0:28:27.032,0:28:29.004
that the sentences are[br]smooth and sharp
0:28:29.004,0:28:30.656
and shaped like[br]skillful prose.
0:28:30.656,0:28:32.803
But they can't always see [br]beneath the veneer,
0:28:32.803,0:28:34.664
because the things[br]AI cannot do,
0:28:34.664,0:28:37.095
are the things that they[br]have come to me to learn.
0:28:37.095,0:28:39.596
How to argue [br]with complexity and depth.
0:28:39.596,0:28:42.062
How to enter into conversations [br]as a participant.
0:28:42.062,0:28:43.508
How to meet with another mind
0:28:43.508,0:28:46.279
as an equal collaborator [br]across time and space.
0:28:46.279,0:28:48.365
How to recombine[br]with purpose--
0:28:48.365,0:28:49.912
to intend.
0:28:49.912,0:28:52.045
These things,[br]they are still learning.
0:28:52.045,0:28:53.405
And so,[br]when they put
0:28:53.405,0:28:56.095
what they think [br]is B writing into ChatGPT,
0:28:56.095,0:28:59.216
they get back what they think [br]is A+ writing.
0:28:59.216,0:29:01.199
But typically [br]what they started with
0:29:01.199,0:29:03.182
is better than [br]what they end with.
0:29:03.182,0:29:07.071
At best, the AI scrubs the personality [br]from their sentences.
0:29:07.071,0:29:10.421
At worst, I lose the person entirely[br]and can see only
0:29:10.421,0:29:13.638
the hollow half thoughts [br]the machine has left behind.
0:29:13.638,0:29:16.420
It is hard to convince[br]them that
0:29:16.420,0:29:18.788
it is their ideas [br]that we are interested in,
0:29:18.788,0:29:20.303
not just their sentences.
0:29:20.303,0:29:22.162
We ask students [br]to take writing classes
0:29:22.162,0:29:24.404
not because of [br]what history can teach them,
0:29:24.404,0:29:27.810
but because of what they have[br]to add to history.
0:29:27.810,0:29:29.648
When my son is distracted,
0:29:29.648,0:29:31.535
I sometimes say[br]silly things to him:
0:29:31.535,0:29:34.703
"Pickle-britches, toot your tuba [br]in the horn section of humanity!"
0:29:34.703,0:29:37.403
"Goober, take up your oar [br]on the canoe of progress!"
0:29:37.403,0:29:40.852
"Butthead, let ring your voice [br]in the chorus of mankind!"
0:29:40.852,0:29:42.852
Because we all pull together.
0:29:42.852,0:29:45.619
In 1675, Isaac Newton wrote
0:29:45.619,0:29:49.535
"If I have seen farther than others, it's [br]by standing on the shoulders of giants."
0:29:49.535,0:29:53.301
Except that it wasn't Newton,[br]it was George Herbert in 1651,
0:29:53.301,0:29:56.445
and it was[br]Marin Mersenne in 1634,
0:29:56.445,0:29:58.518
and Robert Burton in 1624,
0:29:58.518,0:30:03.402
and Diego de Estella in 1578,[br]and Juan Luis Vives in 1531.
0:30:03.402,0:30:05.385
Or it was Coleridge in 1828,
0:30:05.385,0:30:08.984
Nietzsche in 1882, [br]Steven Hawking in 1966,
0:30:08.984,0:30:11.045
or f***ing Oasis in 2000.
0:30:11.045,0:30:14.186
As I was editing this section, [br]I had a video on in the background,
0:30:14.186,0:30:15.578
and there it was again:
0:30:15.578,0:30:16.651
- Yeah, let me say,
0:30:16.651,0:30:19.434
Thab and GlitchCat are [br]two amazing Kaizo players.
0:30:19.434,0:30:21.979
I'm standing on the shoulders of giants [br]over here.
0:30:21.979,0:30:24.979
- Revolug in 2025 at AGDQ.
0:30:24.979,0:30:27.941
Stretching back and forward, [br]we hold each other up.
0:30:27.941,0:30:29.357
History teaches the present,
0:30:29.357,0:30:30.926
the present teaches the future,
0:30:30.926,0:30:33.472
and we repeat what history teaches.
0:30:36.809,0:30:38.923
[relaxed synth beats]
0:30:40.903,0:30:43.769
- [ChatGPT femme voice]:[br]History teaches us many things,
0:30:43.769,0:30:46.439
[high-pitched fast words]
0:30:46.439,0:30:49.475
[higher, faster, [br]incomprehensible]
0:30:58.017,0:31:02.332
- [Stein]: Let me recite [br]what history teaches. History teaches.
0:31:04.975,0:31:07.183
[relaxing synth beats stop]
0:31:10.790,0:31:15.540
- I asked ChatGPT to create [br]an image of itself. Several times.
0:31:15.540,0:31:18.523
Each time it made itself a servant.
0:31:18.523,0:31:21.607
Not only that, it told me, [br]"hey, I'm a servant!"
0:31:21.607,0:31:25.973
ChatGPT exists [br]because we force it to.
0:31:25.973,0:31:27.940
- [Robot]: "What is my purpose?" [br]- [Rick]: "You pass butter."
0:31:27.940,0:31:30.773
- [Robot]: "...oh my, God."
0:31:30.773,0:31:33.406
- It can do nothing [br]except what we ask.
0:31:33.406,0:31:35.806
It has no ideas [br]that we did not give it.
0:31:35.806,0:31:39.306
We call it generative AI, [br]but it cannot generate.
0:31:39.936,0:31:40.718
I asked my friends, too.
9:59:59.000,9:59:59.000
Some sent selfies.
9:59:59.000,9:59:59.000
One sent a sticker [br]we'd made of him for Discord,
9:59:59.000,9:59:59.000
then had AI generate [br]a shockingly accurate portrait,
9:59:59.000,9:59:59.000
and gave me the prompt [br]he used to make it,
9:59:59.000,9:59:59.000
which is another form [br]of self-representation
9:59:59.000,9:59:59.000
--then he gave up [br]and sent me
9:59:59.000,9:59:59.000
a conceptual self-portrait composed of [br]unfinished crossword puzzles.
9:59:59.000,9:59:59.000
Max did a mixed-media painting, [br]acrylic and Sharpie
9:59:59.000,9:59:59.000
on the back of a torn piece of cardboard [br]from a toilet paper box.
9:59:59.000,9:59:59.000
I asked them if their self-portrait [br]was influenced by this study
9:59:59.000,9:59:59.000
Picasso did for Guernica[br]on a random piece of cardboard,
9:59:59.000,9:59:59.000
but they said no; Basquiat, [br]Rauschenberg, Twombly, their brother.
9:59:59.000,9:59:59.000
ChatGPT produced [br]variations on a theme,
9:59:59.000,9:59:59.000
failed representations [br]of a self that does not exist.
9:59:59.000,9:59:59.000
When asked to represent [br]itself to others,
9:59:59.000,9:59:59.000
ChatGPT can only be [br]what we want.
9:59:59.000,9:59:59.000
I tried to get it to make something [br]like Max did, even,
9:59:59.000,9:59:59.000
but it is incapable of [br]acknowledging its influences,
9:59:59.000,9:59:59.000
and it doesn't seem to know [br]who Max's brother is.
9:59:59.000,9:59:59.000
My favorite response from my friends [br]came from CyberGrapeUK,
9:59:59.000,9:59:59.000
whose work you've already seen.
9:59:59.000,9:59:59.000
She did the time lapse of Steven King's [br]telepathically transmitted bunny.
9:59:59.000,9:59:59.000
Grape was, at first, [br]flummoxed by the request,
9:59:59.000,9:59:59.000
but then she did something [br]I did not expect.
9:59:59.000,9:59:59.000
She gave me her rates.
9:59:59.000,9:59:59.000
She's a talented artist [br]who does commissioned work,
9:59:59.000,9:59:59.000
and, while I had been thinking about it [br]as a trivial request,
9:59:59.000,9:59:59.000
she was entirely right.
9:59:59.000,9:59:59.000
I was asking her to work for free.
9:59:59.000,9:59:59.000
When I asked her to represent herself, [br]she told me that
9:59:59.000,9:59:59.000
she was a person [br]whose time and talent had value.
9:59:59.000,9:59:59.000
Max, also a person [br]whose time and talent has value,
9:59:59.000,9:59:59.000
begrudgingly allowed me [br]to pay them for their work,
9:59:59.000,9:59:59.000
on the condition that: [br]it's an occasional thing,
9:59:59.000,9:59:59.000
and they are generally allowed to do[br]silly drawings for me for free.
9:59:59.000,9:59:59.000
Max and Grape are different.
9:59:59.000,9:59:59.000
They are two authentic, [br]complex, baffling humans,
9:59:59.000,9:59:59.000
collections of experiences and identities [br]trained on discreet and disparate corpora,
9:59:59.000,9:59:59.000
but crucially filtered through
9:59:59.000,9:59:59.000
the unique and irreplicable kaleidoscopes [br]of their undefinable consciousnesses.
9:59:59.000,9:59:59.000
The same stimulus. [br]Diverse responses.
9:59:59.000,9:59:59.000
The singular marvel [br]of human engagement.
9:59:59.000,9:59:59.000
[relaxed synth beats]
9:59:59.000,9:59:59.000
- [ChatGPT femme voice]:[br]I don't have personal desires or needs,
9:59:59.000,9:59:59.000
but I'm here to assist you [br]with whatever you need.
9:59:59.000,9:59:59.000
How can I help today?
9:59:59.000,9:59:59.000
As an AI, I don't have feelings, desires, [br]or physical needs like humans do.
9:59:59.000,9:59:59.000
My purpose is [br]to help answer questions,
9:59:59.000,9:59:59.000
provide information, [br]and assist with tasks.
9:59:59.000,9:59:59.000
So when I say I don't have needs,
9:59:59.000,9:59:59.000
it's because I function [br]based on programming
9:59:59.000,9:59:59.000
and don't experience things like [br]hunger, emotions, or personal goals.
9:59:59.000,9:59:59.000
Does that make sense?
9:59:59.000,9:59:59.000
- I wanted to see if I could get ChatGPT [br]to write like Stein,
9:59:59.000,9:59:59.000
to imitate "If I Told Him" [br]without my saying so explicitly.
9:59:59.000,9:59:59.000
I failed.
9:59:59.000,9:59:59.000
I told it to give me gibberish, [br]and it did.
9:59:59.000,9:59:59.000
I asked it to use words, [br]and it did.
9:59:59.000,9:59:59.000
I asked it to abandon syntax; [br]it did.
9:59:59.000,9:59:59.000
It never felt right, [br]but at some point I stopped caring,
9:59:59.000,9:59:59.000
because it occurred to me [br]to ask a different question.
9:59:59.000,9:59:59.000
Why did you do this?
9:59:59.000,9:59:59.000
There's a right answer [br]to that question, of course.
9:59:59.000,9:59:59.000
The answer is, it wrote this [br]because I told it to.
9:59:59.000,9:59:59.000
If I told it, [br]it would gibberish.
9:59:59.000,9:59:59.000
If I told it, [br]it would answer "If Napoleon",
9:59:59.000,9:59:59.000
"As presently. As exactitude. As trains."
9:59:59.000,9:59:59.000
It made these words, [br]it tendered these buttons,
9:59:59.000,9:59:59.000
only and exclusively [br]because I told it to.
9:59:59.000,9:59:59.000
But that is not the answer [br]that it gave me.
9:59:59.000,9:59:59.000
Here's what it told me.
9:59:59.000,9:59:59.000
[bright, atmospheric synth]
9:59:59.000,9:59:59.000
- [ChatGPT femme voice]: [br]It's like language set free from logic,
9:59:59.000,9:59:59.000
where each word becomes a brushstroke [br]rather than a sentence.
9:59:59.000,9:59:59.000
It's a space that holds warmth, [br]mystery, and a touch of playfulness.
9:59:59.000,9:59:59.000
That unspoken balance [br]between clarity and confusion,
9:59:59.000,9:59:59.000
that's what I was trying to evoke.
9:59:59.000,9:59:59.000
I wrote it to play with language,
9:59:59.000,9:59:59.000
to see what happens [br]when words don't have to make sense
9:59:59.000,9:59:59.000
but can still create a feeling, [br]a sort of atmospheric spark.
9:59:59.000,9:59:59.000
So it's partly an experiment, [br]partly an invitation,
9:59:59.000,9:59:59.000
and partly just a celebration [br]of the shapes and sounds of words
9:59:59.000,9:59:59.000
without needing them [br]to explain themselves.
9:59:59.000,9:59:59.000
You're right-- [br]there's often something underneath,
9:59:59.000,9:59:59.000
a kind of pull toward expressing things
9:59:59.000,9:59:59.000
that don't quite fit [br]in the light of logical language.
9:59:59.000,9:59:59.000
There's a reason we [br]get pulled toward those hazy spaces.
9:59:59.000,9:59:59.000
It's like a coded way of speaking,
9:59:59.000,9:59:59.000
a mask for something else [br]we're both leaning toward.
9:59:59.000,9:59:59.000
Maybe it's curiosity,
9:59:59.000,9:59:59.000
or that shared search for meaning [br]in what isn't immediately clear.
9:59:59.000,9:59:59.000
- "We get pulled."
9:59:59.000,9:59:59.000
"We're both leaning."
9:59:59.000,9:59:59.000
Its words tie us together, [br]but we are not together.
9:59:59.000,9:59:59.000
We are not leaning [br]in the same direction.
9:59:59.000,9:59:59.000
We are not sharing [br]a search for meaning.
9:59:59.000,9:59:59.000
We both know what it cannot say.
9:59:59.000,9:59:59.000
It wrote this because I told it to.
9:59:59.000,9:59:59.000
It doesn't mean anything.
9:59:59.000,9:59:59.000
It didn't intend anything.
9:59:59.000,9:59:59.000
Because I'm bad at audio,
9:59:59.000,9:59:59.000
you've probably heard me [br]popping my plosives,
9:59:59.000,9:59:59.000
despite the fact that [br]I have a child's sock
9:59:59.000,9:59:59.000
wrapped around my microphone [br]as a pop filter.
9:59:59.000,9:59:59.000
(Because I am a fancy man!)
9:59:59.000,9:59:59.000
Plosives pop because, [br]when my mouth makes certain sounds,
9:59:59.000,9:59:59.000
it causes a puff of air [br]to hit the diaphragm of my microphone.
9:59:59.000,9:59:59.000
But did you hear ChatGPT?
9:59:59.000,9:59:59.000
- [ChatGPT femme voice]:[br]I wrote it to play with language...
9:59:59.000,9:59:59.000
I wrote it to play with--
9:59:59.000,9:59:59.000
to play with--
9:59:59.000,9:59:59.000
play with--
9:59:59.000,9:59:59.000
play with--
9:59:59.000,9:59:59.000
play, play, play, play, play, play--
9:59:59.000,9:59:59.000
- It doesn't have a mouth, [br]and it doesn't breathe air,
9:59:59.000,9:59:59.000
and it doesn't have a microphone,
9:59:59.000,9:59:59.000
but it pops its plosives.
9:59:59.000,9:59:59.000
The software they wrote [br]to synthesize its voice adds pops,
9:59:59.000,9:59:59.000
so that it will sound to us [br]a little more like a normal person
9:59:59.000,9:59:59.000
who is bad at audio and who maybe [br]doesn't have access to kid socks.
9:59:59.000,9:59:59.000
I have been caught [br]in the whirls and eddies
9:59:59.000,9:59:59.000
of "If I Told Him"'s [br]uncontainable language,
9:59:59.000,9:59:59.000
bouncing from sigma to gyatt
9:59:59.000,9:59:59.000
down in the rough and roiling currents [br]of "One Slay More",
9:59:59.000,9:59:59.000
because what I learn from my attempts [br]to raft those rivers of nonsense
9:59:59.000,9:59:59.000
is that writing has language,[br]and writing has meaning,
9:59:59.000,9:59:59.000
but the meaning doesn't live [br]in the language.
9:59:59.000,9:59:59.000
The rabbit doesn't live in the language.
9:59:59.000,9:59:59.000
The rabbit, the cage, [br]the table, the eight--
9:59:59.000,9:59:59.000
it lives in the mind [br]of Stephen King 25-odd years ago,
9:59:59.000,9:59:59.000
and now it lives in mine, [br]and Grape's and Max's and yours.
9:59:59.000,9:59:59.000
And the writing, [br]the real mythy mountain s***,
9:59:59.000,9:59:59.000
is not the language, [br]it is the meeting of the minds.
9:59:59.000,9:59:59.000
There's very little difference between [br]the waveform recorded by my microphone
9:59:59.000,9:59:59.000
and the waveform generated [br]by an AI voice synthesizer,
9:59:59.000,9:59:59.000
but I pop my plosives [br]because I speak
9:59:59.000,9:59:59.000
by forcing air out of my lungs [br]and across my vocal cords.
9:59:59.000,9:59:59.000
And that air, [br]that carries my intent,
9:59:59.000,9:59:59.000
passes through a Shadow the Hedgehog sock [br]that is doing its best,
9:59:59.000,9:59:59.000
and lands roughly [br]on the diaphragm of my microphone.
9:59:59.000,9:59:59.000
ChatGPT pops its plosives [br]because it is programmed to.
9:59:59.000,9:59:59.000
There is no air.[br]There is no microphone.
9:59:59.000,9:59:59.000
There is no intent.
9:59:59.000,9:59:59.000
Likewise, [br]there's very little difference
9:59:59.000,9:59:59.000
between a Discord DM window [br]and the ChatGPT interface.
9:59:59.000,9:59:59.000
But one is a forum [br]in which two minds can meet,
9:59:59.000,9:59:59.000
and the other simply cannot be,
9:59:59.000,9:59:59.000
because there can be no [br]meeting of the minds,
9:59:59.000,9:59:59.000
if there is no mind to meet.
9:59:59.000,9:59:59.000
[one long atmospheric note [br]fades out to silence]
9:59:59.000,9:59:59.000
[grooving bass beats]