The code and data behind xeiaso.net
talks/2024: add AI ethics talk
Signed-off-by: Xe Iaso <me@xeiaso.net>
+758
+4
lume/src/talks/2024/_data.yml
+4
lume/src/talks/2024/_data.yml
+754
lume/src/talks/2024/prepare-unforeseen-consequences.mdx
+754
lume/src/talks/2024/prepare-unforeseen-consequences.mdx
···
1
+
---
2
+
title: "AI: the not-so-good parts"
3
+
date: 2025-01-16
4
+
tags:
5
+
- ai
6
+
- ethics
7
+
- philosophy
8
+
---
9
+
10
+
Hey, if you normally read the written form of my talks, I highly
11
+
suggest watching or listening to the video for this one. The topic I'm
12
+
covering is something I'm quite passionate about and I don't think
13
+
that my tone is conveyed in text the same way it is in voice. If the
14
+
version on XeDN doesn't load for you for whatever reason, please
15
+
[contact me](/contact/) with the output of
16
+
[cdn.xeiaso.net/cgi-cdn/wtf](https://cdn.xeiaso.net/cgi-cdn/wtf) and I
17
+
will figure out what is wrong so I can fix it.
18
+
19
+
You can find the YouTube version of this talk [here](https://youtu.be/EfAjITmLP50?feature=shared).
20
+
21
+
<XeblogVideo path="talks/2024/ai-ethics" />
22
+
23
+
---
24
+
25
+
<XeblogSlide name="2024/ai-ethics/001" essential />
26
+
27
+
Hi, I'm Xe Iaso and before we get started, I want to start by talking
28
+
about what this talk is and is not. This talk isn't going to be the
29
+
kind of high signal AI research that I'd really love to be giving
30
+
right now. This talk is about actions and consequences.
31
+
32
+
<XeblogSlide name="2024/ai-ethics/002" essential />
33
+
34
+
What impacts will our projects have on the real world where people
35
+
have to take objects like this and exchange them for food and shelter?
36
+
37
+
I'm sorry to say that this talk is going to be a bit of a wet blanket.
38
+
I'm so sorry for Yacine because all that stuff with local AI inference
39
+
in browsers was really cool. And that dogfooding of
40
+
[dingboard](https://dingboard.com/) for a presentation about how
41
+
dingboard works was cool as hell.
42
+
43
+
<XeblogSlide name="2024/ai-ethics/004" essential />
44
+
45
+
All the best things in life come with disclaimers, as I'm sure you
46
+
know, and these words are my own. I'm not speaking on behalf of my
47
+
employer, past employers, or if you're watching the recording and I've
48
+
changed employers, any future employers. I am speaking for myself, not
49
+
other people.
50
+
51
+
Before we get into this, let's cover my background, some stuff about
52
+
me, what I do, and all this AI stuff has benefited and harmed me
53
+
personally. As Hai [the organizer of the AI meetup that asked me to
54
+
speak there] mentioned, I'm a somewhat avid blogger. I've only got
55
+
like 400 articles or something. I write for the love of writing and
56
+
I've got like maybe four 3D printed save icons of text available on my
57
+
blog for anyone to learn with any topic from like programming,
58
+
spirituality, semiotics, AI, etc. My writing is loved by the developer
59
+
community and it's the reason why I get hired.
60
+
61
+
<XeblogSlide name="2024/ai-ethics/007" essential />
62
+
63
+
Regardless of anything I say in this talk, please make a blog,
64
+
document what you've learned, document what works, document what
65
+
fails, just get out there and write. You'll get good at it, just keep
66
+
at it. This is genuine advice.
67
+
68
+
<XeblogSlide name="2024/ai-ethics/008" essential />
69
+
70
+
However, as a reward for making my blog a high-quality thing, it's
71
+
part of the ChatGPT training dataset. Somewhere in some data center,
72
+
my blog's information is sitting there tokenized, waiting to get
73
+
massaged into floating point weights by unfeeling automatons used to
74
+
make unimaginable amounts of money that I will never see a penny of.
75
+
This is the punishment I get for pouring the heart, soul and love into
76
+
my craft as a blogger.
77
+
78
+
I get turned into ChatGPT.
79
+
80
+
<XeblogSlide name="2024/ai-ethics/009" essential />
81
+
82
+
Now in our system of law, things are generally lawful unless there's
83
+
some law or precedent that says it's not. At the time of me speaking
84
+
this, we aren't sure if training AI models on copyrighted
85
+
data is fair use or not. The courts and lawmakers need to battle this
86
+
out (if they'll be allowed to because there is a lot of money behind
87
+
the AI industry right now).
88
+
89
+
This is technology that is so new, it's making Bitcoin look like Stone
90
+
Age, 8-bit computing back when you couldn't count above 255 without
91
+
major hacks.
92
+
93
+
<XeblogSlide name="2024/ai-ethics/010" essential />
94
+
95
+
And mind you, I'm just one blogger. I'm just one person. I don't have
96
+
that big of a platform, all things considered. Sure in the genre of
97
+
technology bloggers, I'm probably fairly high up there, but I'm not
98
+
like front page on New York Times big. I'm just a person who likes
99
+
talking about computers and how they should work. I'm just someone
100
+
that gazed into the void too much and now people to pay me to gaze
101
+
into the damn void.
102
+
103
+
<XeblogSlide name="2024/ai-ethics/011" essential />
104
+
105
+
So how do we understand all this?
106
+
107
+
How do we figure out how to peel back all the layers of terminology
108
+
bullshit that keep us from having a clear understanding of what people
109
+
are even saying?
110
+
111
+
If we take all the drama and interplay involved in our society, we can
112
+
boil it down to two basic things, actions and consequences. Actions
113
+
are the things that we do and consequences are the things that result.
114
+
115
+
So let's say you cut a tree down to make a fire, but that tree was
116
+
used by animals to shelter them from the winter and now those animals
117
+
have a harder time finding shelter in the winter.
118
+
119
+
You take actions and something or someone else has to deal with the
120
+
consequences.
121
+
122
+
Most of the time our actions serve to make us better off and shield us
123
+
from the consequences. We see this happen with that tree that got cut
124
+
down. We will see this happen with ChatGPT and we will keep seeing
125
+
this happen time immemorial as society keeps repeating.
126
+
127
+
As exciting as all of this AI technology is, as a science fiction
128
+
writer, I can't help but see the same actions and consequences and
129
+
analyses for how we're using it today.
130
+
131
+
<XeblogSlide name="2024/ai-ethics/016" essential />
132
+
133
+
Now your pitchforks can go down, I see you out there, you holding them
134
+
up, I'm not trying to be a contrarian or decry AI as wrongthink. I've
135
+
been using AI for my own stuff and I genuinely think that there's a
136
+
lot of really exciting things here.
137
+
138
+
I'm mostly worried about how the existing powers that be are going to
139
+
use this surplus of cheap labor and have those actions have massive
140
+
consequences on us all.
141
+
142
+
<XeblogSlide name="2024/ai-ethics/017" essential />
143
+
144
+
One of the things I'm trying to get across here is not all "Capitalism
145
+
bad! Let's get back the bread lines, baby!" There's plenty of places
146
+
to see those arguments and I don't want this to be one of those. I
147
+
more want to inspire you to see what the consequences of your actions
148
+
with AI stuff could be so that we can make the world a more equitable
149
+
place.
150
+
151
+
Of course, this is made even more fun by the concept of unforeseen
152
+
consequences or downstream consequences that you couldn't have
153
+
possibly seen coming when you were experimenting with things.
154
+
155
+
<XeblogSlide name="2024/ai-ethics/018" essential />
156
+
157
+
As an example, for a long time people thought swans were white. Swans
158
+
became symbols of literary purity or something like that and it was so
159
+
common that there was an English idiom of a black swan being an
160
+
impossible thing.
161
+
162
+
As this photo proves, swans can be black.
163
+
164
+
And now the term "black swan event" describes something that should
165
+
have been obvious in hindsight but something that we couldn't possibly
166
+
have foreseen at the time.
167
+
168
+
(Begin sarcastic tone)
169
+
170
+
Just like that unmentionable-on-YouTube viral pandemic that happened a
171
+
few years ago that our society will never really recover from!
172
+
Scientists were warning us for years that we'd be totally screwed by a
173
+
viral pandemic but no, we didn't take them seriously.
174
+
175
+
(End sarcastic tone)
176
+
177
+
<XeblogSlide name="2024/ai-ethics/020" essential />
178
+
179
+
Whenever anyone takes actions and there are consequences or impacts,
180
+
you can usually model them as on yourself, your friends or the world
181
+
at large. I haven't found a good way to model the impact risk of a
182
+
given field very well, but I like triangles so I made this triangle
183
+
called the impact triangle to show what all of the factors in the
184
+
computer science industry are.
185
+
186
+
In terms of access, anybody can become good at coding and start
187
+
working at a company or creating a company to solve a problem that
188
+
they have in their lives. I'm pretty sure that this basic thing, the
189
+
computer industry is open access to anybody is basically why everybody
190
+
in this room is here today.
191
+
192
+
Personally, I'm a college dropout.
193
+
194
+
Without the industry allowing just about anyone to walk in the door
195
+
and start being successful, yeah, I'd still be in the Seattle area
196
+
probably working minimum wage at a fast food place. I wouldn't be able
197
+
to dream of immigrating to Canada and I probably would have never met
198
+
my husband who is so thankfully recording this for me.
199
+
200
+
There's also no professional certification or license required to
201
+
practice computer science or software development or whatever we call
202
+
ourselves now. And basically anybody off the street without
203
+
certification can make an impact on the world scale if they get lucky.
204
+
205
+
And then in terms of limits, our industry measures results in small
206
+
units of times like individual financial quarters. In aggregate, our
207
+
industry only cares about what we do to make the capitalism line go up
208
+
for next quarter and there's no ethical or professional guidelines
209
+
that prevent people from making bad things or even defining what good
210
+
and bad is in the first place. In an ideal world, the thought is that
211
+
the market should sort everything out and realistically, with the GDPR
212
+
and the like, there are some laws that enable, that force people to
213
+
comply but as long as you have good lawyers, you can get away with
214
+
killing murder.
215
+
216
+
For most other professions in the job market, our industry looks
217
+
incredibly reckless. Like, accountants need to be licensed and pass
218
+
certifications. If you want to call yourself a surgeon, you need to
219
+
have surgery practice, you need to have a license in surgery, and you
220
+
need to keep yourself up with the profession.
221
+
222
+
We don't have such barriers to entry.
223
+
224
+
As an example of this, consider Facebook. They have a billion users.
225
+
That is nine significant figures, a billion with a B as in bat. When
226
+
they made Facebook, the thought was that they could make everybody
227
+
better by reducing the social distance and that could make everybody
228
+
like happier and live more fulfilled lives.
229
+
230
+
An unimaginable amount of photos, video and text posts are made to
231
+
Facebook every day. Some measurable fraction of these violate
232
+
Facebook's community guidelines and are full at the very least and are
233
+
fully legal at the most. Many trivial cases can be handled by machine
234
+
learning algorithms but there's always that bit that needs to be
235
+
judged by a human.
236
+
237
+
Speaking as a recovering IRC op, content moderation is impossible at
238
+
small scales and the level of impossibility only grows as the number
239
+
of people involved in a thing grows. I am fairly certain that it is
240
+
like actually entirely impossible to moderate Facebook at this point
241
+
because there's just too many people. You have to have some machine
242
+
algorithm in there at some point and there are going to be things that
243
+
the algorithm can't handle.
244
+
245
+
So then you go and you use humans to rate that.
246
+
247
+
You contract out a company who very wisely decides to subcontract that
248
+
out because they don't have to deal with the fallout and finally it
249
+
ends up on the desks of people that are tortured day and night by the
250
+
things they are forced to witness to make rent.
251
+
252
+
For the action of creating Facebook and all of the systems that let
253
+
Mark Zuckerberg make a bunker on Hawaii, raise his own cattle, make
254
+
his own beer, and smoke those meats, he doesn't have to see those
255
+
images and things that the content moderators have to see.
256
+
257
+
He just lays back and watches his bank account number go up and maybe
258
+
does CEO things if he has to.
259
+
260
+
The human cost is totally discounted from the equation because the
261
+
only limit is what makes the capitalism line go up. The people doing
262
+
the actions almost never see the consequences because the CEO of Uber
263
+
never got his job replaced by an Uber driver. The CEO of Google never
264
+
suffered the algorithm locking him out of his entire digital life for
265
+
good with no way to get it all back. And the people doing the actions
266
+
and making the decisions are not affected by any of the consequences,
267
+
foreseen or unforeseen.
268
+
269
+
The last time I spoke here, I spoke about a work of satire called
270
+
[Automuse](/videos/2023/ai-hackathon/). Automuse is a tool that uses
271
+
large language models to recreate the normal novel writing process
272
+
using large language models and a good dose of stochastic randomness
273
+
to make some amusing outputs.
274
+
275
+
When I made it, I really just wanted to throw ink to the canvas to see
276
+
what would happen, then write [a satirical scientific
277
+
paper](https://cdn.xeiaso.net/file/christine-static/video/2023/ai-hackathon/automuse-2.pdf).
278
+
279
+
<XeblogSlide name="2024/ai-ethics/031" essential />
280
+
281
+
To my horror, I won the hackathon with a shitpost about the publishing
282
+
industry that was inspired my fear of what could happen if things like
283
+
Automuse were more widespread.
284
+
285
+
When I gave my talk at the hackathon, I had a five minute slot and
286
+
there was something that I had on my script that I cut out as I was
287
+
speaking.
288
+
289
+
Not sure why I did, it just felt right at the time.
290
+
291
+
The part that I left out was inspired by this quote from the
292
+
philosopher SammyClassicSonicFan:
293
+
294
+
<XeblogSlide name="2024/ai-ethics/033" essential />
295
+
296
+
When will you **learn**?
297
+
When will you learn that your **actions** have **consequences**?
298
+
299
+
I made Automuse precisely because I understand how impractical such a
300
+
thing is. The output quality of Automuse will never compare to what a
301
+
human can write no matter what large language model you throw at it.
302
+
303
+
Okay, yes. I did my research, there's actually a rather large market
304
+
for low quality pleasure reading that something like Automuse could
305
+
fill. There's a surprisingly large number of people that enjoy reading
306
+
formulaic things about good winning out over evil or old people
307
+
reading romance novels to feel the passion of being young again or
308
+
whatever. Not to mention doing something like that as a company would
309
+
leave me an excellent moat because most AI companies want to focus on
310
+
the high quality super output and here I am, the trash vendor going
311
+
in, yeah, I'd basically be invincible.
312
+
313
+
But I don't know if I could live with myself if I turned Automuse
314
+
into a product.
315
+
316
+
When I made Automuse, I knew that this was a potentially high impact
317
+
thing, so I crippled it.
318
+
319
+
I made it difficult for anyone to use, even me.
320
+
321
+
I made it rely on a private NPM dependency that is on a server that
322
+
only I have the API token to and it just so happens to be the thing
323
+
that generates random plots.
324
+
325
+
I also made it in a way that requires massive human intervention and
326
+
filtering in order to get decent results and every so often I get a
327
+
message from somebody that asks me:
328
+
329
+
<BlockQuote>Hey, how can I set up Automuse on my stuff?</BlockQuote>
330
+
331
+
And they're surprised when I quote them a five figure number to get
332
+
them to go away. And some are even angry and curse me out because a
333
+
person making open source software on the internet would want to be
334
+
paid for their time.
335
+
336
+
I can't understand that actually.
337
+
338
+
But above all, the reason why I really don't want to productize it or
339
+
make it available for mass consumption in any form is the problem of
340
+
book spam. Automuse would make the problem of book spam worse.
341
+
342
+
The Book Spam problem is where people upload nonsense to the Kindle
343
+
store and make boatloads of money doing it. This problem has been
344
+
accelerated by ChatGPT and is getting to the point where Amazon's book
345
+
vending thing actually had to implement rate limits for uploading
346
+
books.
347
+
348
+
I don't think I could live with myself if I made and released an easy
349
+
to use product that made that problem worse.
350
+
351
+
It's bad enough that whenever I get around to finishing my novel
352
+
Spellblade (I couldn't find the cover I commissioned, so I just put
353
+
the name on the slide), I'm almost certainly just going to release it
354
+
on itch.io or to my patrons for very cheap. In theory, the Kindle
355
+
store would be the best place for that kind of high signal original
356
+
fiction but I just don't want it to get flooded out in a wave of AI
357
+
generated mushroom foraging books.
358
+
359
+
I don't think that anyone at OpenAI anticipated that people would use
360
+
ChatGPT to make the book spam problem worse. I have a friend that
361
+
works there and generally from what I've seen, the research side of
362
+
OpenAI really has their head screwed on the right way.
363
+
364
+
The problem is the capitalism side of OpenAI getting that sweet, sweet
365
+
return to an investment by making a product that nobody else can
366
+
provide and then charging for the output.
367
+
368
+
<XeblogSlide name="2024/ai-ethics/039" essential />
369
+
370
+
Above all, the part that really confuses me is why we're automating
371
+
away art and writing instead of like snow blowing or something
372
+
actually useful. There's a subtle part of me that's really concerned
373
+
for the future of our industry and I really think we need to be aware
374
+
of it before it all bites us and like getting rid of everybody that
375
+
has aesthetic knowledge really seems like a bad idea for an industry
376
+
that focuses so much on design.
377
+
378
+
<XeblogSlide name="2024/ai-ethics/040" essential />
379
+
380
+
With the Industrial Revolution came factories. Factories allowed us to
381
+
produce objects on scales like never before. Raw materials go in at
382
+
one end, human labor goes in the middle, finished products come out
383
+
the end. This has allowed us to become the kind of species we are
384
+
today. You can circumnavigate the globe in 100 hours while playing a
385
+
contrived game show about travel. You can head to an entirely
386
+
different continent in like what, 12 hours and this has led us to
387
+
discoveries that have made us healthier, lived longer lives and
388
+
overall it's been a boon for the human race.
389
+
390
+
<XeblogSlide name="2024/ai-ethics/041" essential />
391
+
392
+
However, this is a modern assembly line for cars. Look what you don't
393
+
see here, people. All of those robot arms and the like represent jobs
394
+
that were done by humans, operating the crane to lower the truck body
395
+
onto the chassis, all of that stuff. With every new model year there's
396
+
more automation at play and less room for human jobs.
397
+
398
+
Sure, we can make more cars per hour but like every job that's not
399
+
done by a human is another family that can't make rent. It's another
400
+
child that can't grow up and you know actually cure cancer or
401
+
something. And I just feel like it's another way for the ownership
402
+
class to scrape more off the top.
403
+
404
+
With that in mind, I want you to consider this:
405
+
406
+
<XeblogSlide name="2024/ai-ethics/042" essential />
407
+
408
+
These are our factories, the open office environment. Instead of wool
409
+
or wood or water as input, we have user stories, electricity and
410
+
coffee. Many of the companies out there are really just assembly lines
411
+
for code features or Kubernetes configurations. I think the ultimate
412
+
dream of this lies in the idea of the T-shaped developer that I've
413
+
seen many management people talk about when they're trying to
414
+
reorganize their companies.
415
+
416
+
<XeblogSlide name="2024/ai-ethics/043" essential />
417
+
418
+
The core idea of the T-shaped developer is that you have really good
419
+
competency in one field and enough broad knowledge in other fields
420
+
that you can basically be put anywhere in a project and be useful.
421
+
This is why you see things like ephemeral teams or decrees from on
422
+
high that thou must write in JavaScript for all things.
423
+
424
+
And in theory, it makes it a lot easier to move people around and
425
+
place them wherever the company needs in order to make the process
426
+
more adaptable to the circumstances. Not to mention, if everyone's
427
+
just a T-shaped developer, that makes it really easy to get people off
428
+
of the street and into the job in days so you don't have to spend the
429
+
months training them on how you messed up Jenkins this time.
430
+
431
+
Ever notice that every job opportunity is only for senior roles?
432
+
433
+
This is why.
434
+
435
+
Usually by the time you convince companies to give you a title that
436
+
starts with the word "Senior", you've already been molded into a
437
+
T-shaped engineer and you can slot in just about anywhere.
438
+
439
+
This is our assembly line, created in the fear that if we don't do
440
+
this, the wrong line will trend in the wrong way and investors won't
441
+
give us as much money as freely.
442
+
443
+
Like, okay, I realize I'm doing some doom and gloom stuff here.
444
+
445
+
It's probably going to be a while until AI is actually able to replace
446
+
our jobs. Right now, there isn't a magic button that product teams can
447
+
use to "just implement that feature" based on the textual description.
448
+
That's probably a long ways off and it'll probably require a different
449
+
fundamental architecture than attention window transformer models.
450
+
451
+
But with that in mind, there's a segment of people that already have
452
+
the magic "just implement it" button today:
453
+
454
+
Artists.
455
+
456
+
Stable diffusion, mid-journey, and Dall-E 3 have gotten to the point
457
+
where the output is not just good.
458
+
459
+
It's good enough.
460
+
461
+
For the vast majority of people, as long as there's nothing obviously
462
+
wrong with the hands, you won't be able to tell an image that is AI
463
+
generated.
464
+
465
+
However, artists can tell instantly when you have an AI generated
466
+
illustration.
467
+
468
+
<XeblogSlide name="2024/ai-ethics/049" essential />
469
+
470
+
Just look at this one I used earlier in this talk. It's so bad. Look
471
+
at the stem on that flower. That is not how stems work. The brush at
472
+
the bottom is just blending into the easel in ways that physically
473
+
separate objects don't work. The flower that the robot is holding is
474
+
inconsistent. It looks like the light is coming from both forward and
475
+
backward at the same time. The antennae are melting into the shoulders
476
+
of the robot.
477
+
478
+
It's totally passable at first glance.
479
+
480
+
I'm pretty sure that before I mentioned all those stuff and put all
481
+
the arrows on the slide, you wouldn't have seen any of it. But when
482
+
you start critically analyzing it, it just falls to pieces.
483
+
484
+
I guess the better question here is why would you want to use an AI
485
+
generated image for something?
486
+
487
+
One of the big places you want to use an AI image is for the cover
488
+
image on your blog post because we've come to expect that blog posts
489
+
need cover images for some reason.
490
+
491
+
There's more desire for people to have cheap filler art that meets a
492
+
certain criteria than there are artists willing to work for
493
+
unrealistically low prices with incredibly quick turnaround times. Art
494
+
is everywhere and yet it's commoditized so much that it's worthless in
495
+
a day and age where rent and food prices keep going up.
496
+
497
+
So we end up with something like this:
498
+
499
+
<XeblogSlide name="2024/ai-ethics/051" essential />
500
+
501
+
You get an AI generated of assembly line of robots painting flowers.
502
+
503
+
This is really why I didn't want to develop Automuse into a company. I
504
+
just fear that action would have too many consequences and my friends
505
+
and fellow artists would suffer. This is why I did so much detailed
506
+
math about how much it would cost per word, how the quality would be
507
+
seen in the market, and what impact such a technology would have if it
508
+
churned out hundreds of books per hour.
509
+
510
+
Outside of the systems we live in, yeah, this AI stuff is great. It's
511
+
fantastic tech that allows us to do any number of things we couldn't
512
+
do before.
513
+
514
+
But inside the systems we live in, I can't say the help, let's see
515
+
this is yet another way that human labor is being displaced without a
516
+
good replacement.
517
+
518
+
And we wonder why we can't call ourselves engineers in Ontario. Do we
519
+
really engineer anything or are we just making the line go up?
520
+
521
+
When will we learn that our actions have consequences?
522
+
523
+
Until then I guess we need to prepare for unforeseen consequences.
524
+
525
+
Thank you all for watching this and I hope it gives you some things to
526
+
think about. I hope I didn't break too many taboos about the industry
527
+
in the process but who am I kidding? I just broke all of them.
528
+
529
+
<XeblogSlide name="2024/ai-ethics/061" essential />
530
+
531
+
Thanks to everyone on this list for inspiring me to take action and
532
+
pushing towards the presentation I gave tonight. Special thanks to
533
+
Mystes and Layl for really grinding hard into this, ripping in half
534
+
and telling me where I'm full of shit. Extra special thanks to my
535
+
husband for recording this for me and thank you for watching.
536
+
537
+
<XeblogSlide name="2024/ai-ethics/062" essential />
538
+
539
+
I recognize that this is like really a heavy talk. It'll probably take
540
+
you some time to surface some good questions about it but if you
541
+
happen to have them right now please feel free to ask. I will be happy
542
+
to answer but if it takes you a while to come up with it just email
543
+
[unforeseenconsequences@xeserv.us](mailto:unforeseenconsequences@xeserv.us).
544
+
It'll get to my inbox and I promise you I will reply. Have a good
545
+
evening and does anyone have any questions?
546
+
547
+
## Q&A
548
+
549
+
<BlockQuote>What was the sigil you displayed at the beginning of your
550
+
talk?</BlockQuote>
551
+
552
+
That was the sigil of Baphomet, one of the names for Satan as
553
+
celebrated in Satanism.
554
+
555
+
<BlockQuote>Do you see a future where AI technology can equitably help
556
+
humanity thrive?</BlockQuote>
557
+
558
+
I do see a future where it can be used to benefit us all. The problem
559
+
is the intersection of what could be, what is, and the tools in the
560
+
process where you get the real interesting stuff and there's probably
561
+
at least five good sci-fi novels you could write about this.
562
+
563
+
You could write a really compelling one about just what happened with
564
+
OpenAI and especially what's happened with the e/acc people. I wrote
565
+
the plot outline for a bad science fiction novel about the madness
566
+
that is e/acc.
567
+
568
+
<BlockQuote>What do you think we should do about this
569
+
problem?</BlockQuote>
570
+
571
+
Just be aware that your actions don't exist in a vacuum.
572
+
573
+
If you build something that could replace jobs, then you need to be
574
+
cognizant of the people that you're going to make unable to pay rent
575
+
because if you make something that replaces knowledge work labor, you
576
+
price them out of being able to eat. When people aren't able to afford
577
+
to eat, they especially can't afford to retrain themselves to work in
578
+
another industry that hasn't been taken over by infinite cheap labor.
579
+
580
+
<BlockQuote>
581
+
First, thank you very much for the presentation. I'm not debating
582
+
here. I'm very open for these type of discussions, but you show the
583
+
industrial revolution and the next slide was all the people who were
584
+
impoverished. I don't see it as a linear change though, so industrial
585
+
revolution and all those workers working in those situations by itself
586
+
was not a necessary, better situation than those workers in those
587
+
dangerous situations being replaced by robots on the other side. As we
588
+
move on, we never had any occasions that we needed to get rid of a
589
+
bunch of populations because we didn't have jobs for them, but we
590
+
eventually came up with solutions, new jobs, some sort of a solution.
591
+
So the main question is how do you see that change exactly from
592
+
industrial revolution to industrial revolution?
593
+
</BlockQuote>
594
+
595
+
At some level, this stuff is going to happen regardless, and if it's
596
+
going to happen, there should be some societal support mechanism, like
597
+
universal basic income (which no matter what study is made to prove it
598
+
doesn't work, actually does work) to replace the income that we're
599
+
losing to machines taking over jobs that were previously done by
600
+
humans. Something like universal basic income would probably help a
601
+
lot here, but I don't know.
602
+
603
+
I don't have any solutions.
604
+
605
+
I'm more trying to blow the whistle that there's a problem before it
606
+
gets bad enough that things become irreparable.
607
+
608
+
<BlockQuote>
609
+
All right, I'd like to commend you first on your courage to do this.
610
+
It's obviously difficult to come into a room and say the opposite. At
611
+
the same time, I'll give you the opposite and the pit that was out of
612
+
the pit. You know, one of the things that, to act your way a little
613
+
bit, automation is known to increase the standard of living. So we
614
+
have all great things we can do because of automation. So AI is
615
+
automation's superpower. Now to say there's no consequences of AI
616
+
being abused, there definitely will be, but looking at the greater
617
+
impact of it all, and I think that's the reason we're at all here, is
618
+
because we know that they're [unintelligible], but truly down, we know
619
+
that bringing abundance to the world is far greater and needs to be
620
+
substantial in that event.
621
+
</BlockQuote>
622
+
623
+
I mean, yes, congratulations. You actually got the point of the talk.
624
+
The point of the talk is to get you to think critically about what
625
+
these tools are, what's going on, and what the benefits could be as
626
+
well as what the downsides could be. I just don't know if our current
627
+
system of distributing wealth and resources is really going to be able
628
+
to adapt to that in time without some major cataclysm forcing the
629
+
measure.
630
+
631
+
<BlockQuote>
632
+
I just wanted to ask you.
633
+
You said you're not sure if this system of wealth distribution is the
634
+
right system that should be, you know, that should have this kind of
635
+
AI in place for moving forward. So what kind of system do you think is
636
+
more practical for that?
637
+
</BlockQuote>
638
+
639
+
640
+
So I think one of the more ideal outcomes would be if people that
641
+
whose work is in the training set of ChatGPT end up getting royalties
642
+
from OpenAI for their data being used to make unimaginable amounts of
643
+
money.
644
+
645
+
Like, I have been transformed into ChatGPT. I can't go back to college
646
+
because all of my writing comes back as flagged by AI because I've
647
+
written so much and it's in so many different data sets that it just
648
+
keeps getting flagged as AI generated.
649
+
650
+
And like, yeah, we all know the AI generation plagiarism checkers are
651
+
bullshit and people shouldn't use them yet the colleges do for some
652
+
reason.
653
+
654
+
So like, what can you do?
655
+
656
+
Really the best possible way to get equity here would be to basically
657
+
make it so that if you research AI with copyrighted materials, that's
658
+
fine. But when it comes to putting the money generator in the mix,
659
+
hold up, maybe you actually need to pay royalties because those blog
660
+
posts and the like, they don't just come out of nowhere for free.
661
+
Like, you know, you have to train to be an artist. Like, this photo of
662
+
this log that I got off of Pexels, a public domain image stock image
663
+
sit, you have to have some like skill in photography to know the rule
664
+
of thirds and you know, like be able to configure your camera to
665
+
capture the exact moment of this log falling like this. There are
666
+
actual skills that don't look like skills that still require a lot of
667
+
time, energy, and frankly, remuneration to compensate for.
668
+
669
+
I think one of the best ways would be to make the concept of an open
670
+
source model that is only just the weights without any of the training
671
+
data or any of the training methodology involved an unterm.
672
+
673
+
Like, that is not open source, that is open access. Open source would
674
+
be providing all of the code you used for training, all of the data
675
+
that you used for training, and a summary of where you got the data
676
+
from.
677
+
678
+
That would be closer to what open source actually is and anything
679
+
close to the definition of open source back when the GPL was the
680
+
dominant definition of open source.
681
+
682
+
Generally open source AI stuff is really cool. There's a lot of stuff
683
+
you can do with it. I'm just really concerned about the intersection
684
+
between that and, you know, the capitalism system that we're all
685
+
forced to live under.
686
+
687
+
<BlockQuote>How do we combat abuse or data that isn't labeled as AI
688
+
generated? Are we in the death of the Information Age because of
689
+
this?</BlockQuote>
690
+
691
+
Oh. I have no idea.
692
+
693
+
On my blog I've been tracking AI-generated content farms and the tools
694
+
that they use to do it because it's kind of horrifying how easy it is
695
+
to get ChatGPT to hallucinate something about how to make soap with
696
+
radishes.
697
+
698
+
By the way, don't do that. It'll kill you. It will actually kill you
699
+
dead. Do not do that. No, I'm actually serious here.
700
+
701
+
The worst part is how this intersects with content farms, those random
702
+
websites you find on Google with negative amounts of information and
703
+
ads everywhere. I've already seen ChatGPT make that problem worse.
704
+
705
+
Hell, there was this SEO heist a while ago where this person basically
706
+
fed Google Trends results into ChatGPT, SEO heisted by rewriting their
707
+
competitor's website entirely from scratch and stole all their traffic
708
+
and made a whole bunch of ad money contributing nothing to society.
709
+
710
+
I don't really know how this is all going to work out, but I really
711
+
hope we're not in the death of the information age because that's what
712
+
pays my bills. But if things keep going the way they're going, I can't
713
+
help but agree that we may be on the decline of everything getting
714
+
drowned in pages of trivia and celebrity bullshit.
715
+
716
+
<BlockQuote>
717
+
Thanks for talking. You see that, you know, that the technology
718
+
naturally democratizes people's access to information. Won't more
719
+
access to information make things better for everyone?
720
+
</BlockQuote>
721
+
722
+
I'm very glad that inference is getting so much cheaper, like, hell,
723
+
this MacBook right here (I would lift it up, but it's hooked up via
724
+
USB and I don't want to disconnect it). It can run Mixtral [a model
725
+
considered roughly equivalent to GPT-3.5, the model used for ChatGPT]
726
+
and it's just a random MacBook off the shelf.
727
+
Looking back, I kind of regret not getting as much RAM because I
728
+
didn't think I would be doing all this, but, you know, c'est la vie
729
+
[Canadian idiom meaning "that's life"].
730
+
731
+
I have been thinking about doing an experiment of using Q-LoRA to
732
+
train the ultimate recommendation engine based off of posts that I've
733
+
either commented on or upvoted on Hacker News and using that as input
734
+
with the classification of like or dislike. And because I downvote or
735
+
flag a fair number of posts there, I can use that to create a somewhat
736
+
rough aggregate of things that I would be interested in. And that
737
+
would be something that I see could be a really interesting
738
+
application of all this.
739
+
740
+
Like I said, though, the open source AI stuff is really cool, but the
741
+
intersection between that and the system and the powers that be today,
742
+
I don't know how that's going to happen and I'm just afraid that it
743
+
won't end up good for all of us.
744
+
745
+
But thank you for all the questions. I am really happy that I was able
746
+
to get you to be engaged with this topic and really start thinking
747
+
because I don't know what's going to happen either.
748
+
749
+
Thank you so much. Good night all! Drive home safely! The roads are
750
+
wild.
751
+
752
+
753
+
754
+