When Should We Recognize Something as a Property Right?

WEBVTT
1
00:00:02.080 --> 00:00:05.080
Welcome to FEDSOC Forums, a podcast of the Federal Societies
2
00:00:05.120 --> 00:00:08.279
Practice Groups. I'm niy kas Merrit, Vice President and director
3
00:00:08.320 --> 00:00:11.320
of Practice Groups at the Federal Society. For exclusive access
4
00:00:11.320 --> 00:00:14.640
to live recordings of fedsock Forum programs, become a Federal
5
00:00:14.679 --> 00:00:17.160
Society member today at fedsoc dot org.
6
00:00:18.440 --> 00:00:21.039
Good afternoon, everyone, and welcome to fedzoc Forums.
7
00:00:21.280 --> 00:00:22.920
My name is Lipy Dickinson and I'm.
8
00:00:22.760 --> 00:00:25.839
An associate director with the Federalist Societies Practice Groups. As
9
00:00:25.879 --> 00:00:28.480
a reminder, all opinions expressed are those of our speakers
10
00:00:28.480 --> 00:00:31.480
and not of the Federalist Society. Today's program will be
11
00:00:31.519 --> 00:00:34.280
a discussion of when should we recognize something as a
12
00:00:34.320 --> 00:00:39.520
property rate. Today's moderator is Satya Marar, postgraduate research fellow
13
00:00:39.520 --> 00:00:42.920
with the Mercada Center at George Mason University. If our
14
00:00:42.960 --> 00:00:45.399
audience has any questions throughout the webinar that pertain to
15
00:00:45.439 --> 00:00:48.200
the topic at hand, please submit them through Zoom's Q
16
00:00:48.320 --> 00:00:50.520
and a feature located at the bottom of your screen
17
00:00:50.520 --> 00:00:52.799
and not the chat, so we'll be able to access
18
00:00:52.840 --> 00:00:55.399
them later. With that note, I will turn things over
19
00:00:55.479 --> 00:00:58.159
to our speakers to start the conversation. Satcha, thank you
20
00:00:58.159 --> 00:01:00.000
so much for joining us today and for moderating today
21
00:01:00.159 --> 00:01:00.799
is discussion.
22
00:01:01.719 --> 00:01:02.320
Thanks for that.
23
00:01:03.119 --> 00:01:06.200
As we approach it two fifty anniversary of the American Republic,
24
00:01:06.400 --> 00:01:09.640
we recall an America's constitutional experiment began with one of
25
00:01:09.719 --> 00:01:12.959
many striking propositions that the new Republic will protect not
26
00:01:13.000 --> 00:01:15.959
only land, ships, tools, and goods, but also products of
27
00:01:16.000 --> 00:01:19.920
the mind. Article one, Section eight clauset of the Constitution
28
00:01:20.000 --> 00:01:22.480
gives Congress the power to promote the progress of science
29
00:01:22.519 --> 00:01:25.920
and the useful arts by securing for limited time, exclusive
30
00:01:26.000 --> 00:01:29.079
rights for authors and inventors in their writings and discoveries.
31
00:01:29.719 --> 00:01:32.280
This is a remarkable clause, not merely because it authorized
32
00:01:32.319 --> 00:01:36.200
copyright and patterns, but because it embeds IP not simply
33
00:01:36.239 --> 00:01:39.040
as the policy proposition, but within a constitutional theory of
34
00:01:39.120 --> 00:01:43.359
ordered liberty, republican self government, and economic diynrism. It treats
35
00:01:43.400 --> 00:01:47.159
creativity in invention as public goods worth encouraging, while also
36
00:01:47.200 --> 00:01:50.560
limiting the duration and purpose and exclusive rights. The idea
37
00:01:50.640 --> 00:01:53.079
isn't that ideas themselves should be owned the way you
38
00:01:53.120 --> 00:01:55.959
own a farm or a house. It's a subtler premise.
39
00:01:56.359 --> 00:02:00.879
That's a society that wants learning, invention, expressive culture, industrial progress,
40
00:02:00.879 --> 00:02:04.799
and technological diffusion must sometimes grant legal enforceable rights or
41
00:02:04.879 --> 00:02:11.479
intangible assets. Those rights channel private initiative towards publicans, giving authors, inventors, firms,
42
00:02:11.520 --> 00:02:13.120
investors enough control.
43
00:02:12.800 --> 00:02:14.560
To risk time and capital.
44
00:02:15.120 --> 00:02:18.919
They allow trade, licensing, specialization and markets to flourish. They
45
00:02:18.960 --> 00:02:22.319
create incentives to disclose rather than to conceal useful ideas.
46
00:02:22.840 --> 00:02:26.639
But they also carried injurers, monopoly, rent seeking, suppression of
47
00:02:26.719 --> 00:02:30.120
flow on innovation, and overclaiming of what belongs within the
48
00:02:30.120 --> 00:02:35.240
common stock knowledge. The modern debate over AI data markets
49
00:02:35.280 --> 00:02:37.759
and so on have returned us back to a deeper
50
00:02:37.840 --> 00:02:41.560
question than IP. When should the law recognize something as
51
00:02:41.560 --> 00:02:44.639
a property right. The answer can't simply be whenever something
52
00:02:44.680 --> 00:02:48.639
as value. Facts, ideas, language, scientific truth, that human style
53
00:02:48.719 --> 00:02:52.039
all have value, but they can't all be enclosed. Not
54
00:02:52.120 --> 00:02:55.000
can the answer be only when the exclusion costs nothing,
55
00:02:55.039 --> 00:03:00.159
Because every serious property regime imposes some social costs. One
56
00:03:00.199 --> 00:03:04.919
main questions these situations seem to come down to, is
57
00:03:04.919 --> 00:03:09.960
the right to exclude necessarily useful to induce valuable creation, investment, coordination,
58
00:03:10.520 --> 00:03:13.759
disclosure of personal interests. Are the boundaries the property right
59
00:03:13.840 --> 00:03:17.080
that we're going to create administrable by courtes, tribunals and regulators.
60
00:03:17.400 --> 00:03:20.039
Will a new right increase or reduce transaction costs on
61
00:03:20.120 --> 00:03:24.199
netterms or change the transaction cost burden unduly in favor
62
00:03:24.199 --> 00:03:26.280
of one side or the other? And would giving the
63
00:03:26.319 --> 00:03:29.479
rights holder this degree of control? What will the implication
64
00:03:29.560 --> 00:03:33.680
of that be for free speech, competition, science, democratic culture,
65
00:03:33.879 --> 00:03:37.159
any number of policy considerations. To help us answer and
66
00:03:37.199 --> 00:03:40.319
clarify our current debates around expanding or scaling back the
67
00:03:40.319 --> 00:03:45.039
scope of property rights, we have today three great experts. First,
68
00:03:45.120 --> 00:03:48.000
we have alder Abbott, senior Research fellow at the Market
69
00:03:48.159 --> 00:03:51.479
Center and George Mason University and a former general counsel
70
00:03:51.639 --> 00:03:55.840
at the FTC, Professor Jen Bambauer of the University of
71
00:03:55.840 --> 00:03:59.319
Florida where she teaches law on journalism. And Jeffrey deb
72
00:04:00.120 --> 00:04:02.879
So for Law Policy and the Committee for Justice. I'm
73
00:04:02.879 --> 00:04:05.360
a moderator post graduate research fellow, and look at is
74
00:04:05.719 --> 00:04:12.080
sofia mr r So. With that, let's kick things off, Jane,
75
00:04:12.719 --> 00:04:15.599
what makes property rights different from other kinds of rights,
76
00:04:15.639 --> 00:04:18.120
such as on the taught contract or statute.
77
00:04:19.279 --> 00:04:21.600
Yeah, I've given this a lot of thought because even
78
00:04:21.639 --> 00:04:25.040
though I cannot claim to be a property theorist or
79
00:04:25.079 --> 00:04:27.079
really even a property law scholar, and I'm going to
80
00:04:27.399 --> 00:04:30.839
rely on Jeffrey and Alden to help help us, you know,
81
00:04:31.000 --> 00:04:34.360
kind of elevate the discussion.
82
00:04:34.519 --> 00:04:35.240
On that side.
83
00:04:35.519 --> 00:04:39.560
I have found that my writing about privacy law and
84
00:04:39.600 --> 00:04:45.800
about free speech, the conception of information as property has been,
85
00:04:46.079 --> 00:04:48.759
you know, has for decades now been sort of creeping
86
00:04:48.800 --> 00:04:53.639
in and causing kind of some tension and some some
87
00:04:53.759 --> 00:04:58.000
kind of intellectual fuzziness in in in these areas. So
88
00:04:58.000 --> 00:05:00.040
so I think what's key to property in this This
89
00:05:00.160 --> 00:05:03.120
is debatable among property scholars, but I think it's it's
90
00:05:03.199 --> 00:05:08.319
clear that the main key right involved with property is
91
00:05:08.680 --> 00:05:11.279
the right to exclude. So if somebody has and it
92
00:05:11.279 --> 00:05:14.079
doesn't have to be an absolute right to exclude, but
93
00:05:14.079 --> 00:05:17.519
if someone has the power in most circumstances to decide
94
00:05:17.560 --> 00:05:20.360
who gets to use a resource and who doesn't, and
95
00:05:20.439 --> 00:05:24.879
if the courts will give them compensation when when their
96
00:05:25.079 --> 00:05:29.079
wishes are violated, irrespective of having to prove fault or
97
00:05:29.160 --> 00:05:32.040
kind of societal harm, then you know that you have
98
00:05:32.079 --> 00:05:34.639
a property right rather than some other type of right.
99
00:05:34.920 --> 00:05:39.079
Contrast that with things like contract law, where a violation
100
00:05:39.240 --> 00:05:42.800
requires an agreement in advance between people. So it's more
101
00:05:42.839 --> 00:05:46.160
relational and contrast that with tort law, as you said,
102
00:05:46.480 --> 00:05:48.879
or risk management law doesn't necessarily have to be common
103
00:05:48.920 --> 00:05:54.000
law tort, but whether it's public statutes or torts, if
104
00:05:54.040 --> 00:05:57.519
the ideas that well, we're setting in advance some expectations
105
00:05:57.519 --> 00:06:01.199
about types of risks we don't want to imposed on society,
106
00:06:01.759 --> 00:06:05.920
then we would ask, did someone else's use of this
107
00:06:06.120 --> 00:06:08.399
resource cause harm?
108
00:06:08.439 --> 00:06:08.560
You know?
109
00:06:08.680 --> 00:06:09.079
Was it was?
110
00:06:09.079 --> 00:06:09.839
It?
111
00:06:09.959 --> 00:06:13.160
Was it fault blameworthy for the other person to have
112
00:06:13.240 --> 00:06:18.040
used it. So so using this, you know, I think
113
00:06:18.040 --> 00:06:21.240
this applies very easily to things that have a thinginess
114
00:06:21.319 --> 00:06:23.879
to them. You know, So if I'm if I have
115
00:06:23.959 --> 00:06:26.319
a car, it's my car, and other people can't use
116
00:06:26.319 --> 00:06:28.160
my car because then I wouldn't be able to use
117
00:06:28.160 --> 00:06:29.639
and I wouldn't be able to sort of.
118
00:06:29.759 --> 00:06:31.279
Get the best value out of it.
119
00:06:32.000 --> 00:06:36.720
But obviously with non rivalrous and tangible goods like information
120
00:06:37.000 --> 00:06:41.120
or ideas or content of any sort, it gets a
121
00:06:41.120 --> 00:06:41.959
little trickier.
122
00:06:42.560 --> 00:06:42.720
Uh.
123
00:06:42.759 --> 00:06:45.680
There are still some good reasons to assign property rights,
124
00:06:46.319 --> 00:06:48.160
but there are also some good reasons, as you said
125
00:06:48.199 --> 00:06:52.439
in your introductory remarks, to be cautious, because by assigning
126
00:06:52.480 --> 00:06:55.759
property rights, you're setting this default that a single person
127
00:06:55.800 --> 00:06:58.600
gets to decide what they think the best use is
128
00:06:58.680 --> 00:07:02.079
of that item, and it may be that may be
129
00:07:02.519 --> 00:07:04.680
sort of wrong, and maybe that other people actually have
130
00:07:04.759 --> 00:07:08.240
a much better idea of what to do with that resource,
131
00:07:08.360 --> 00:07:12.360
and maybe their interests either conflict or a sort of
132
00:07:12.439 --> 00:07:14.120
orthogonal to the original rights holder.
133
00:07:15.800 --> 00:07:18.720
Cool Cool all than Jess, do you have anything to
134
00:07:19.000 --> 00:07:21.040
add to that definition of those distinctions?
135
00:07:25.160 --> 00:07:28.000
I think I think Jane summarizing very well. I mean,
136
00:07:29.319 --> 00:07:33.160
the key thing about property, someone gets tricky because there's
137
00:07:33.199 --> 00:07:36.040
a moral dimension, don't they talk about that a little
138
00:07:36.040 --> 00:07:40.480
bit later? As well as a purely utilitarian dimension. It's
139
00:07:40.519 --> 00:07:44.319
not just what what rules allow the maximum output of
140
00:07:44.399 --> 00:07:48.399
a certain go to goods or services or expressions.
141
00:07:48.600 --> 00:07:51.639
They're all for societies also, are you know?
142
00:07:51.720 --> 00:07:54.759
How are we are we able to protect the rights
143
00:07:55.279 --> 00:07:59.560
of individuals to protect what they have, what they have
144
00:07:59.680 --> 00:08:02.480
created because they don't want others to use it.
145
00:08:02.639 --> 00:08:04.639
The right to exclude is and what was saying Princess.
146
00:08:04.680 --> 00:08:09.040
A trademark owner may have a specific market, identifies and
147
00:08:09.120 --> 00:08:12.160
prevents confusion and good or service, and thus may not
148
00:08:12.240 --> 00:08:16.360
want anyone else to use that. There may be an
149
00:08:16.360 --> 00:08:19.879
economic reason for that if others use it but degrade
150
00:08:19.920 --> 00:08:24.399
the quality. It can cause confusion, it may lose its value.
151
00:08:24.399 --> 00:08:27.120
But at the same time it's also there's a moral dimension.
152
00:08:27.160 --> 00:08:29.680
I came up with the idea and the thoughts to
153
00:08:30.480 --> 00:08:33.840
use this, and I want to control it completely. That's
154
00:08:33.919 --> 00:08:37.159
particularly the case in the area so called rights and personality,
155
00:08:37.200 --> 00:08:40.519
which have existed for about the century or so and
156
00:08:40.600 --> 00:08:44.080
half essentially mainly in state common law and same state's
157
00:08:44.159 --> 00:08:47.480
tangritory law. But they're becoming very important right now. Think
158
00:08:47.559 --> 00:08:52.879
of the college athletes playing football and baseball. Now they
159
00:08:52.960 --> 00:08:57.799
are being paid, usually by sports supporters of particular school,
160
00:08:58.279 --> 00:09:01.639
large amounts of money to attend this school, and supposedly
161
00:09:01.720 --> 00:09:04.559
as to protect your name, image, and likeness. The idea
162
00:09:04.840 --> 00:09:09.399
is that if an individual has invested talents in developing
163
00:09:09.399 --> 00:09:13.960
a particular personal as Michael Jackson or particular athlete, they
164
00:09:14.000 --> 00:09:17.399
have some right to control it. But there are always
165
00:09:17.519 --> 00:09:20.320
boundary questions. You know, how far does this right extent?
166
00:09:21.080 --> 00:09:24.559
Just as how soili fide trademark. Going back to trademark,
167
00:09:24.600 --> 00:09:27.480
how far does that extent? Yeah, it may be, it
168
00:09:27.519 --> 00:09:30.159
may be right. You have a right to protect what
169
00:09:30.240 --> 00:09:33.440
you develop, but what if you don't police it? For instance,
170
00:09:33.440 --> 00:09:38.000
the term kleenex, a term Sermus used to or trademarks
171
00:09:39.039 --> 00:09:45.200
that referred to specific inventions by particular parties. But they
172
00:09:45.240 --> 00:09:48.000
didn't check, they didn't try to prevent other people from
173
00:09:48.120 --> 00:09:51.720
using the term generically, and then and then they lost
174
00:09:51.759 --> 00:09:55.320
control over it. The idea is, this is certain rights
175
00:09:55.399 --> 00:10:00.399
you need to maintain control over. And if you owned
176
00:10:01.360 --> 00:10:04.759
the claim that you are preventing confusion and so on,
177
00:10:05.000 --> 00:10:08.440
or preventing the or have a right to prevent the
178
00:10:08.559 --> 00:10:12.559
benefits of which generic use fall away. So it's tricky,
179
00:10:13.039 --> 00:10:16.919
and really these questions differ upon it based upon the
180
00:10:17.000 --> 00:10:20.279
type of right you're dealing with. For example, copyright doesn't
181
00:10:20.879 --> 00:10:25.240
protect ideas. There's many ideas that are very similar about
182
00:10:25.279 --> 00:10:28.120
a particular novel John gone with them when the pride
183
00:10:28.120 --> 00:10:36.480
and prejudice, But it protects particular manifestations, particular expressions, very
184
00:10:36.519 --> 00:10:42.399
precise expressions of those ideas. So I think I'll stop there.
185
00:10:42.919 --> 00:10:46.559
So there are lots of tricky boundary questions and laws.
186
00:10:46.639 --> 00:10:49.799
I think you can already see in designing whether property
187
00:10:49.879 --> 00:10:54.759
right is being properly applied in case or not.
188
00:10:56.240 --> 00:10:57.279
Yeah, I think it's.
189
00:10:57.159 --> 00:10:59.720
Also something really important that right because we often talk
190
00:10:59.759 --> 00:11:04.480
about these things in very kind of utilitarian terms. You know,
191
00:11:04.519 --> 00:11:07.039
we need to have these rights so we can encourage
192
00:11:07.120 --> 00:11:10.840
investment and economic growth and all that good stuff. But
193
00:11:10.879 --> 00:11:14.559
there's also moral dimension, you know, whether it's fairness sometimes
194
00:11:14.559 --> 00:11:18.600
an amorphice concept giving people control over their own reputations
195
00:11:19.559 --> 00:11:22.120
and the value that they bring to you know, whatever
196
00:11:22.159 --> 00:11:28.120
gets generated. So, yeah, fascinating stuff, Jane. Some of your
197
00:11:28.200 --> 00:11:33.320
research has focused on, you know, calls for the right
198
00:11:33.360 --> 00:11:35.960
to publicity, you know, artists who want more control over
199
00:11:36.000 --> 00:11:39.559
their imagure reputation. Uh, those of us who want more
200
00:11:39.720 --> 00:11:42.039
control over our own data that we generate when we
201
00:11:42.080 --> 00:11:45.080
go onto social media platforms and so on, given how
202
00:11:45.399 --> 00:11:48.879
large and profitable these companies are. Uh, there's bleeds and
203
00:11:48.919 --> 00:11:53.679
do talk about data privacy rules and sometimes which which
204
00:11:53.679 --> 00:11:58.039
sometimes seem to treat our data as a quasi privacy right.
205
00:11:59.399 --> 00:12:00.559
So could you you could.
206
00:12:00.519 --> 00:12:03.480
Comment a little bit about you know, legal movements in
207
00:12:03.480 --> 00:12:06.519
this area, whether here or in Europe. And you know,
208
00:12:06.559 --> 00:12:08.919
if not through these sorts of quasi property rights and
209
00:12:09.000 --> 00:12:13.159
give people control, how else might we address these concerns?
210
00:12:15.000 --> 00:12:18.679
Yeah, like picking up on something else and said, it's
211
00:12:18.679 --> 00:12:23.039
true that copyright law does not cover ideas. There needs
212
00:12:23.080 --> 00:12:26.519
to be a particular, you know, particularized expression that in
213
00:12:26.639 --> 00:12:29.519
order to be copyrightable. But these other areas of law
214
00:12:29.679 --> 00:12:35.080
are not quite so strict. And so if if the
215
00:12:35.120 --> 00:12:38.840
idea is, well, the fact that I visited this particular
216
00:12:38.879 --> 00:12:41.600
website and clicked on this you know, I'm thinking about
217
00:12:41.639 --> 00:12:46.200
web browsing data right clicked on this particular link, that is,
218
00:12:46.679 --> 00:12:48.879
you could treat it as like a fact or an
219
00:12:48.919 --> 00:12:52.320
idea or something that wouldn't be copyrightable. And yet nevertheless,
220
00:12:53.080 --> 00:12:56.840
for kind of instinctive, whether whether we call it moral
221
00:12:57.039 --> 00:13:00.440
or you know, something else, and the instinct think is
222
00:13:00.480 --> 00:13:03.080
for people to say, hey, that's yeah, it's information, but
223
00:13:03.120 --> 00:13:07.679
it's my information. And and so a lot of privacy
224
00:13:07.799 --> 00:13:11.919
law it didn't quite start out this way, but uh,
225
00:13:12.399 --> 00:13:16.919
for the last few decades at least, especially with you know, Europe,
226
00:13:17.480 --> 00:13:21.039
i'd say being at the forefront, there's been an eagerness
227
00:13:21.080 --> 00:13:25.240
to try to recognize through law that that very instinct
228
00:13:25.960 --> 00:13:28.440
that even though this is just a fact, I mean,
229
00:13:28.440 --> 00:13:30.399
you could you could treat the right of personality the
230
00:13:30.840 --> 00:13:32.840
sort of the same way, like this, this is what
231
00:13:33.120 --> 00:13:35.279
my face looks like. I'm a celebrity or you know,
232
00:13:35.320 --> 00:13:37.320
I'm a college football player.
233
00:13:38.480 --> 00:13:40.039
This is the sort of movement I make.
234
00:13:40.360 --> 00:13:44.159
You know, It's not like it's not like these games
235
00:13:44.200 --> 00:13:47.279
necessarily replicate exactly what they did in a particular game,
236
00:13:47.360 --> 00:13:53.600
But the idea is taking on some information that that
237
00:13:53.759 --> 00:13:57.559
the you know, the that the privacy kind of victim
238
00:13:57.600 --> 00:14:01.240
feels like should have been under their control. And so
239
00:14:01.440 --> 00:14:05.039
Europe at least has tried to, you know, through through
240
00:14:05.320 --> 00:14:10.279
the GDPR and other sort of antillary uh laws, has
241
00:14:10.759 --> 00:14:15.759
has tried to kind of maximize that control model. There
242
00:14:15.799 --> 00:14:19.000
are other ways that a business, there are other justifications
243
00:14:19.000 --> 00:14:21.960
for a business to use personal data, but those other
244
00:14:22.120 --> 00:14:27.120
justifications are interpreted extremely narrowly, which means that getting consent
245
00:14:27.440 --> 00:14:30.279
is the way that, you know, most of the time
246
00:14:30.399 --> 00:14:34.039
that the company will be able to do something unusual
247
00:14:34.120 --> 00:14:38.919
or new right new with the data. And so I've
248
00:14:39.200 --> 00:14:42.639
come out sort of critical of this in part because
249
00:14:42.720 --> 00:14:47.679
every time, uh, you know, every time there's a property interest,
250
00:14:47.799 --> 00:14:49.799
whether it's sounding in you know, sort of moral or
251
00:14:49.840 --> 00:14:53.799
economiconomic theory, uh, every time that someone's making acclaim to
252
00:14:53.840 --> 00:14:57.679
a property interest in some piece of information, it is
253
00:14:57.759 --> 00:15:02.080
carving out some liberty interest in someone that someone else
254
00:15:02.240 --> 00:15:06.120
might claim, you know, maybe even with with the attribute
255
00:15:06.159 --> 00:15:08.159
of you know, sort of moral the moral high ground.
256
00:15:09.759 --> 00:15:14.879
So if if I cannot observe or record, or share
257
00:15:15.039 --> 00:15:19.639
or subsequently use some information about how people browse the
258
00:15:19.679 --> 00:15:24.919
internet or in order to you know, produce new knowledge
259
00:15:25.399 --> 00:15:28.759
about the world, then then that is in some sense
260
00:15:29.360 --> 00:15:35.120
taking away an opportunity from from the users of the
261
00:15:35.240 --> 00:15:39.200
data that might claim to have a liberty interest.
262
00:15:41.240 --> 00:15:41.440
Yeah.
263
00:15:41.759 --> 00:15:46.480
One thing also is a good point Jane made is data.
264
00:15:46.600 --> 00:15:51.960
For example, what what if people voluntarily provide, say their
265
00:15:52.000 --> 00:15:56.639
birthday to address some other personal data. Lots of it
266
00:15:56.679 --> 00:16:01.320
actually maybe discoverable on on online and my third parties,
267
00:16:01.360 --> 00:16:04.480
but they're just using it to have access to certain database.
268
00:16:04.799 --> 00:16:07.679
And let's say the users collect that and many other people,
269
00:16:08.440 --> 00:16:12.320
many other individuals who are browsing anonymoum anonymize it, but
270
00:16:12.440 --> 00:16:16.039
allow that to sort of direct programs people of a
271
00:16:16.039 --> 00:16:19.600
certain aide or background, which may actually benefit them.
272
00:16:19.480 --> 00:16:21.960
More and they may be willing to You.
273
00:16:21.840 --> 00:16:24.600
Can also direct ads in a certain way. So in
274
00:16:24.639 --> 00:16:28.240
other words, the value has been created. Yes, this information
275
00:16:28.840 --> 00:16:33.440
quote unquote belongs to a particular person or but it
276
00:16:33.480 --> 00:16:36.440
is it's a third party is created value. So then
277
00:16:36.480 --> 00:16:39.639
you get in a question about it has this value creation,
278
00:16:41.320 --> 00:16:45.879
how do you weigh that against any privacy rise, particularly.
279
00:16:45.279 --> 00:16:48.200
When in the online world.
280
00:16:48.440 --> 00:16:52.679
In the online world, many examples of supposing historically non
281
00:16:52.679 --> 00:16:56.960
public information may be easily obtainable, because if you create
282
00:16:57.039 --> 00:17:02.080
too much of a blockage, then valuable applications which could
283
00:17:02.080 --> 00:17:07.559
benefit a lot of people consumers and businesses online can't
284
00:17:07.599 --> 00:17:12.000
be developed because of legal threats of legal liability.
285
00:17:13.680 --> 00:17:15.359
Yeah, I think one problem is that data.
286
00:17:15.640 --> 00:17:19.799
You know, the person described in the data does not
287
00:17:21.079 --> 00:17:25.759
have a good sense of the likely consequences, both positive
288
00:17:25.759 --> 00:17:28.279
and negative. And so this is where I think there
289
00:17:28.319 --> 00:17:31.599
is a deviation from at least the economic utilitarian theory
290
00:17:31.640 --> 00:17:34.279
of property, where normally we think the person with the
291
00:17:34.359 --> 00:17:36.160
resource they have a lot of you know, this is
292
00:17:36.279 --> 00:17:37.960
like the high X model where they have a lot
293
00:17:38.039 --> 00:17:42.880
of inside information that and so if they if they
294
00:17:43.240 --> 00:17:45.880
know the value and they know the value of giving
295
00:17:45.880 --> 00:17:48.839
it away as well, that's not as true or at
296
00:17:48.880 --> 00:17:51.359
least I think the transaction costs the information costs of
297
00:17:51.359 --> 00:17:55.319
figuring out what your life is going to be like
298
00:17:55.359 --> 00:17:58.279
if you do or do not let let companies track you.
299
00:17:58.920 --> 00:18:02.440
It's very hard for consumers or really anyone to know
300
00:18:02.880 --> 00:18:06.559
with great certainty, but people do tend to undervalue what
301
00:18:06.640 --> 00:18:09.880
they get back, as Alden said, from from some of
302
00:18:09.920 --> 00:18:13.000
these tracking activities, Yeah, I.
303
00:18:12.920 --> 00:18:14.839
Think, I mean those are really interesting points.
304
00:18:14.839 --> 00:18:17.160
And also I mean, you know, you touched on something
305
00:18:17.160 --> 00:18:20.440
important there, which is the implicit assumption that the property
306
00:18:20.519 --> 00:18:22.160
rights sold will know how to.
307
00:18:22.200 --> 00:18:23.279
Best use the property.
308
00:18:23.680 --> 00:18:25.319
I can think of a number of examples where, you know,
309
00:18:25.359 --> 00:18:29.039
maybe that isn't literally true, right, you know, where you know,
310
00:18:29.079 --> 00:18:31.920
often people who own a property will kind of I
311
00:18:31.920 --> 00:18:34.640
mean they own a share in it so on, through
312
00:18:34.720 --> 00:18:37.839
maybe an intermediary, and then they pay for those services
313
00:18:37.880 --> 00:18:40.559
for it to be managed their asset management firms and
314
00:18:40.559 --> 00:18:42.920
so on. I mean to put your point though, with
315
00:18:42.960 --> 00:18:45.359
regard to data, I mean, I think there is an
316
00:18:45.440 --> 00:18:47.839
argument to be made that the economic utility of the
317
00:18:47.880 --> 00:18:51.920
data is in most cases not in that individual piece
318
00:18:51.920 --> 00:18:55.480
of data itself. You know, the way like a drop
319
00:18:55.519 --> 00:18:58.079
of oil is one small share in a big barrel
320
00:18:58.119 --> 00:19:01.480
of oil, right, it seems to be more so aggregating
321
00:19:01.519 --> 00:19:04.799
it and turning it into something that has utility, which
322
00:19:04.880 --> 00:19:08.640
in most cases what the platform would normally be doing.
323
00:19:08.759 --> 00:19:11.079
And of course you know, there isn't that same economic
324
00:19:11.160 --> 00:19:14.480
incentive to for someone to pay for someone to have
325
00:19:14.480 --> 00:19:19.039
the best use of their data. That's seems to me anyway,
326
00:19:19.079 --> 00:19:20.240
But thanks a lot for that, you know.
327
00:19:20.279 --> 00:19:23.039
One of the interesting things also that you mentioned.
328
00:19:26.440 --> 00:19:28.680
Property rights often come down to, you know, is there
329
00:19:28.680 --> 00:19:32.039
a thinkiness the thing if it's tangible or intangible, And
330
00:19:32.079 --> 00:19:35.960
some intangible things are thinking this that others don't. I'm
331
00:19:36.000 --> 00:19:38.440
reading now, you know, some articles arguing that we should
332
00:19:38.480 --> 00:19:41.920
change our copyright law and recognize corporate in artistic style
333
00:19:42.359 --> 00:19:45.920
because AI tools have now allowed, you know, you to
334
00:19:45.920 --> 00:19:49.440
create an algorithm that can arguably quantify that the style
335
00:19:49.480 --> 00:19:52.079
of a drake or a tailors with which I find interesting.
336
00:19:52.079 --> 00:19:54.720
I'm not personally convinced, but you know, there are some
337
00:19:54.839 --> 00:19:56.920
interesting movements in the space.
338
00:19:58.440 --> 00:20:02.039
With that in mind, I want to move on to a.
339
00:20:02.000 --> 00:20:06.759
Controversial topic that's being widely discussed nowadays, the use of
340
00:20:07.079 --> 00:20:11.559
the creational generative AI tools and training them often using
341
00:20:12.119 --> 00:20:15.279
you know, the sum total of human knowledge, including copyrighted works.
342
00:20:15.839 --> 00:20:20.759
So well, for Jeff quote unquote information he owns to
343
00:20:20.799 --> 00:20:24.759
be free, right. AI tools are often just, some would argue,
344
00:20:24.759 --> 00:20:28.359
more efficient to doing what humans do learning patterns. Similarly,
345
00:20:28.400 --> 00:20:32.599
in forod producing creative outputs now we hold humans liable
346
00:20:32.839 --> 00:20:36.119
for violating copyright if they release art, music, of books
347
00:20:36.160 --> 00:20:40.319
that replicate the existing works that permission not for studying.
348
00:20:39.839 --> 00:20:41.279
And learning works.
349
00:20:41.640 --> 00:20:45.200
So why should we treat AI tools differently by extending
350
00:20:45.200 --> 00:20:47.200
copyrights to the training.
351
00:20:50.720 --> 00:20:51.880
Well, you know, I.
352
00:20:54.880 --> 00:20:59.559
Struggle with them a lot of the current discussion on
353
00:20:59.599 --> 00:21:02.759
the top because it's it's always limited to a sort
354
00:21:02.759 --> 00:21:07.920
of a descriptive or positive perspective, And to me, the
355
00:21:07.920 --> 00:21:10.640
better discussion is one is more of a normative perspective.
356
00:21:10.640 --> 00:21:13.799
In other words, you know, I don't necessarily disagree that
357
00:21:13.880 --> 00:21:19.400
under the current sort of fair use regime that you
358
00:21:19.400 --> 00:21:22.400
you can certainly make a strong argument that you know
359
00:21:23.519 --> 00:21:29.039
that this is fair use. But again my question, again
360
00:21:29.119 --> 00:21:32.000
is from a different perspective, is should it be you
361
00:21:32.200 --> 00:21:37.559
should we be utilizing copyrighted works? Should we be utilizing
362
00:21:38.079 --> 00:21:42.279
the works of authors, you know, without compensation?
363
00:21:43.319 --> 00:21:43.759
Uh?
364
00:21:43.799 --> 00:21:47.359
And it implicates you know, a number of concerns I have,
365
00:21:47.680 --> 00:21:51.720
particularly in the I P space, about you know, the
366
00:21:51.839 --> 00:21:55.960
separation of powers issues, uh, and whether you know we
367
00:21:55.960 --> 00:21:58.279
should be leaving so much of this to the course.
368
00:21:59.519 --> 00:22:02.359
You know, the Constitution.
369
00:22:02.440 --> 00:22:05.359
Article one, section eight, clause eight gives plenary power to
370
00:22:05.440 --> 00:22:10.279
Congress to set these rules, and yet you know, we
371
00:22:10.440 --> 00:22:13.400
have you know again, and to the point that you
372
00:22:13.720 --> 00:22:17.880
brought up in the copyright, the transformative use training data,
373
00:22:18.240 --> 00:22:23.200
you know, you know, you know, it's all on this
374
00:22:23.319 --> 00:22:28.480
sort of again untransformative use, which is again is entirely
375
00:22:28.519 --> 00:22:30.319
a court created construct.
376
00:22:30.599 --> 00:22:32.319
You know, it was from a you know, it comes
377
00:22:32.359 --> 00:22:33.440
back from their early.
378
00:22:33.279 --> 00:22:36.599
Nineties, from a judge who wrote a law review article,
379
00:22:37.039 --> 00:22:39.839
and it was picked up by Justice suitor in the
380
00:22:39.880 --> 00:22:43.039
Campbell case, and all of a sudden boom, there it is.
381
00:22:43.119 --> 00:22:46.640
It's now where in transformative the word transformative does not
382
00:22:46.680 --> 00:22:48.200
appear in the statute anywhere.
383
00:22:48.440 --> 00:22:52.279
It's entirely a judge made construct. That and what's.
384
00:22:52.079 --> 00:22:55.680
Happened is what always happens when judges in the courts
385
00:22:55.720 --> 00:22:59.039
get involved where they shouldn't be, is it just expands, right,
386
00:22:59.039 --> 00:23:03.400
and now it's just assumed and subsumed the entire you
387
00:23:03.440 --> 00:23:06.319
know debate, uh, you know, And if we just sort
388
00:23:06.319 --> 00:23:09.680
of go back to sort of first principles founding principles,
389
00:23:09.680 --> 00:23:12.559
which you know, that's sort of my perspective, you know,
390
00:23:12.640 --> 00:23:16.039
and why you know, to my answer to your initial question,
391
00:23:16.160 --> 00:23:19.200
the title, the title of the webinar here is you know.
392
00:23:19.240 --> 00:23:22.759
My answer is, you know, early and often, you know
393
00:23:22.880 --> 00:23:25.720
property when I should be found early, early and often.
394
00:23:25.759 --> 00:23:28.680
I'm the school of the student of the Austrian school,
395
00:23:28.759 --> 00:23:32.119
and I think you know, you know that kind of
396
00:23:32.240 --> 00:23:35.720
you know, uh, you know is essential to store the
397
00:23:35.759 --> 00:23:39.880
market to create value. Is creating those property rights expansively
398
00:23:40.240 --> 00:23:44.279
and early and simply and allowing the market process to
399
00:23:44.319 --> 00:23:47.279
take it to do what it does, and allow parties
400
00:23:47.319 --> 00:23:49.839
to adapt. And I just I don't think we're sort
401
00:23:49.839 --> 00:23:52.200
of giving ourselves enough opportunity. But again, getting back to
402
00:23:52.319 --> 00:24:00.960
again the transformative copyright case again, in my view, you know, uh,
403
00:24:01.160 --> 00:24:05.880
you know, the proper analysis again is going back to
404
00:24:05.920 --> 00:24:08.240
the statute. Whether the statute talks about the nature and
405
00:24:08.279 --> 00:24:11.079
the various factors, the nature and purpose of the use. Well,
406
00:24:11.319 --> 00:24:13.839
this is as commercial use as you can possibly get.
407
00:24:14.000 --> 00:24:15.799
But we never seem to talk about.
408
00:24:15.480 --> 00:24:18.079
That in the discussion of whether this is fair use.
409
00:24:18.160 --> 00:24:21.480
It always jumps to the judge made doctrine of is
410
00:24:21.519 --> 00:24:24.480
it transformative or not? You know, My view is, you know,
411
00:24:24.720 --> 00:24:25.799
that's not the way to go.
412
00:24:26.480 --> 00:24:26.799
Uh.
413
00:24:26.839 --> 00:24:31.480
And we're depriving these authors, these innovators, these rights holders,
414
00:24:31.519 --> 00:24:36.680
of proper compensation for their work and Frankly, I think
415
00:24:36.720 --> 00:24:39.160
in the long run, we're harming ourselves because if you
416
00:24:39.279 --> 00:24:41.519
find a way to compensate these people the way that
417
00:24:41.599 --> 00:24:44.039
you know, based on the way the system is set up,
418
00:24:44.400 --> 00:24:46.759
you're going to get more of what they do, and
419
00:24:46.799 --> 00:24:49.000
that's going to benefit all of us. And that's going
420
00:24:49.039 --> 00:24:52.160
to benefit you know, the AI industry or you know,
421
00:24:52.480 --> 00:24:52.960
and this.
422
00:24:52.920 --> 00:24:54.640
Applies in the patent space as well.
423
00:24:55.240 --> 00:24:58.240
You're going to get more uh, you know, innovation, which
424
00:24:58.279 --> 00:25:01.160
is again what we all say we all want.
425
00:25:01.759 --> 00:25:01.960
Uh.
426
00:25:02.039 --> 00:25:04.440
And if they're properly compensated, they're going to do more work,
427
00:25:04.440 --> 00:25:06.279
they're going to generate more and that's going to benefit
428
00:25:06.359 --> 00:25:08.599
where we need to go and again and and and
429
00:25:08.720 --> 00:25:13.279
you know, I have no desire to to hold back
430
00:25:14.279 --> 00:25:18.039
the AI industry. I believe the wholeheartedly that we need
431
00:25:18.079 --> 00:25:19.839
to win this race. We need to be the leader,
432
00:25:20.599 --> 00:25:23.240
and we need to innovate there. But there ought to
433
00:25:23.279 --> 00:25:26.799
be a way to respect these folks property rights in
434
00:25:26.839 --> 00:25:30.720
a way again that again respects them and incentivizes them
435
00:25:30.759 --> 00:25:33.200
to do more, which is going to be feed for
436
00:25:33.279 --> 00:25:35.640
a better outcome on on the on the a on
437
00:25:35.720 --> 00:25:39.279
the users of this data, the large language model creators,
438
00:25:39.680 --> 00:25:43.119
and uh you know and and so I think that again,
439
00:25:43.839 --> 00:25:47.640
if we set some simple categorical rules here, the parties
440
00:25:47.680 --> 00:25:50.559
will adapt. We have seen many license agreements take place
441
00:25:50.599 --> 00:25:52.960
for the use of this data, so it.
442
00:25:52.920 --> 00:25:53.759
Can be done.
443
00:25:54.160 --> 00:25:56.119
And as I always say, we found a way to
444
00:25:56.160 --> 00:25:59.400
compensate you know, a recording artist when their songs are
445
00:25:59.400 --> 00:26:01.640
played ringed on the Internet. We can figure out a
446
00:26:01.680 --> 00:26:04.880
way for some reasonable compensation to authors. Again, they're not
447
00:26:04.880 --> 00:26:07.039
going to get rich off of this, but it will
448
00:26:07.079 --> 00:26:10.839
show that we're respecting their property rights and it will
449
00:26:10.880 --> 00:26:13.839
create certainly no disincentive to do more, which is I
450
00:26:13.839 --> 00:26:16.640
think is what we what we really want out of
451
00:26:16.720 --> 00:26:19.319
the system here. So you know that that's kind of
452
00:26:19.440 --> 00:26:22.880
you know where I think about anyway, you know, sort
453
00:26:22.880 --> 00:26:26.680
of the you know, the use of copyrighted materials in
454
00:26:27.119 --> 00:26:28.920
the large language model training setting.
455
00:26:29.920 --> 00:26:31.480
Right, can I sue?
456
00:26:32.240 --> 00:26:34.119
So go ahead, dow then you.
457
00:26:34.079 --> 00:26:37.880
Know, the Supreme Court in twenty twenty one and sixty
458
00:26:38.000 --> 00:26:44.000
two said the role that Google is compying of eleven thousand,
459
00:26:44.119 --> 00:26:47.680
five hundred lines of job by api I'll developed by
460
00:26:47.759 --> 00:26:52.119
Oracle was cop fair use because it was transformative allowed
461
00:26:52.200 --> 00:26:55.039
Google programmers to use some of your skills to create
462
00:26:55.119 --> 00:26:59.200
new smartphone natural you know, they could have easily entered
463
00:26:59.240 --> 00:27:02.319
into a license land Google, and there were a number
464
00:27:02.319 --> 00:27:05.200
of commentators have said there's a huge amount of copyright
465
00:27:05.680 --> 00:27:09.279
or competitive here. This is no lesson theft and calling
466
00:27:09.359 --> 00:27:12.440
it a transformative use. This was just a tiny bit.
467
00:27:12.759 --> 00:27:15.440
This was a huge amount, and it created a huge
468
00:27:15.480 --> 00:27:20.079
value for Google. But oracles and lots of share and sharing.
469
00:27:19.799 --> 00:27:25.039
That bad, you know, So I might come around eventually
470
00:27:25.079 --> 00:27:28.200
in a second to agree with you both, and especially
471
00:27:28.200 --> 00:27:30.880
you Jeff, but I actually still though I'm struggling a
472
00:27:30.920 --> 00:27:31.720
little bit with the.
473
00:27:34.119 --> 00:27:37.200
Conceptually with this. It's if you're talking.
474
00:27:36.880 --> 00:27:39.240
Purely about cases that could be brought against an AI
475
00:27:39.279 --> 00:27:44.759
company based on output that is that is uh, you know,
476
00:27:44.920 --> 00:27:48.039
whether transformative or not, that is that is a copy,
477
00:27:48.559 --> 00:27:50.960
then everything you said makes sense to me. But a
478
00:27:51.039 --> 00:27:52.920
lot of these claims are not about the output. They're
479
00:27:52.920 --> 00:27:56.279
about the input of training the model. And in that sense,
480
00:27:56.319 --> 00:27:57.960
if we're kind of going to, you know, do our
481
00:27:58.000 --> 00:28:02.519
analogy to the humans, it's like saying that because I've
482
00:28:02.559 --> 00:28:06.839
read Shakespeare and if I later some you know, in
483
00:28:06.880 --> 00:28:10.720
some even unknown to me way sort of process that
484
00:28:10.880 --> 00:28:15.640
and wound up writing a totally different story about starcross lovers,
485
00:28:15.640 --> 00:28:19.920
but didn't use that phrase that that itself is some
486
00:28:19.960 --> 00:28:22.799
sort of uh is you know, is some sort of
487
00:28:22.880 --> 00:28:24.720
violation of the Coveryright Act And I don't, so I
488
00:28:24.720 --> 00:28:25.160
don't see that.
489
00:28:25.200 --> 00:28:25.720
So I'm wondering.
490
00:28:26.519 --> 00:28:29.880
So what I'm struggling though, with, is that you're right though,
491
00:28:29.960 --> 00:28:33.200
that if there's no recognition of the value of input
492
00:28:33.240 --> 00:28:36.279
training data, of the you know, hold rights holders for
493
00:28:36.480 --> 00:28:40.160
for input training data. Thinking about the New York Times
494
00:28:40.160 --> 00:28:42.720
case in particular, it means that all of this work,
495
00:28:42.759 --> 00:28:45.720
all of this expensive work that they do with investigative journalism,
496
00:28:45.759 --> 00:28:49.599
can just be dumped into an LM and the knowledge
497
00:28:49.799 --> 00:28:52.319
is kind of extracted in this, you know, in this
498
00:28:52.640 --> 00:28:55.240
in this sort of mysterious way, and the output will
499
00:28:55.279 --> 00:28:58.039
never be a copy. And so they sort of I
500
00:28:58.279 --> 00:29:01.000
see the incentive problem, but I mean, just a straight
501
00:29:01.079 --> 00:29:04.640
up application of the copyright statute. I don't see how
502
00:29:04.640 --> 00:29:08.440
you get from the statute we have today to to
503
00:29:09.200 --> 00:29:11.720
liability based on input data input only.
504
00:29:14.119 --> 00:29:16.440
I guess my thought is, again, you know, they're clearly
505
00:29:16.480 --> 00:29:18.759
making a copy of the work, which is one of
506
00:29:18.759 --> 00:29:21.519
the rights protected by the statute. And it's certainly the
507
00:29:21.559 --> 00:29:24.319
expression that matters because they create these vectors off of
508
00:29:24.359 --> 00:29:28.720
the actual specific expression that the author used.
509
00:29:28.759 --> 00:29:31.119
That's how the vectors are created. What word order.
510
00:29:31.319 --> 00:29:35.480
So it's to me it's essentially what copyright is. This
511
00:29:36.279 --> 00:29:40.359
is designed to protect and uh, you know, and again,
512
00:29:40.440 --> 00:29:41.240
you know, again I.
513
00:29:41.160 --> 00:29:43.640
Struggle with this whole. You know, the law, as we know,
514
00:29:43.839 --> 00:29:45.240
always lacks right.
515
00:29:45.359 --> 00:29:47.920
So you know, this is a you know, this fair
516
00:29:48.039 --> 00:29:51.759
use has been around for two hundred almost two hundred years.
517
00:29:51.960 --> 00:29:52.599
You know, it was.
518
00:29:52.559 --> 00:29:56.359
Codified in seventy six and then again then this then
519
00:29:56.359 --> 00:30:00.680
the courts got involved and expanded it. U. But you know,
520
00:30:00.799 --> 00:30:04.000
I just you know, I don't think that you know,
521
00:30:04.839 --> 00:30:06.680
applying it as a as a.
522
00:30:06.680 --> 00:30:08.400
You know, a fifty year old statute and a two
523
00:30:08.440 --> 00:30:09.599
hundred year old doctrine.
524
00:30:09.720 --> 00:30:11.720
You know, it needs to be updated, you know, to
525
00:30:12.119 --> 00:30:16.400
you know, to adapt to the current use, uh and
526
00:30:16.680 --> 00:30:20.039
to respect property rights. And again I just I don't
527
00:30:20.079 --> 00:30:22.000
think it is That's why again I really think this
528
00:30:22.079 --> 00:30:22.319
is a.
529
00:30:22.279 --> 00:30:24.960
More of a normative discussion. What should it be?
530
00:30:25.640 --> 00:30:28.279
Uh, you know, and we should not be taking people's
531
00:30:28.359 --> 00:30:34.720
work without compensation. That's just anti the American experiment, our
532
00:30:34.759 --> 00:30:35.839
founding principles.
533
00:30:36.119 --> 00:30:38.160
Back to the Declaration, which again is how I look
534
00:30:38.160 --> 00:30:38.960
at this stuff.
535
00:30:39.599 --> 00:30:43.079
Uh, you know requires that that they did these rights
536
00:30:43.119 --> 00:30:46.640
be be protected, be respected, and be compensated.
537
00:30:47.160 --> 00:30:47.400
Uh.
538
00:30:47.480 --> 00:30:49.799
And you know again any other issue I just want
539
00:30:49.799 --> 00:30:53.279
to point out again it was my concern, my where
540
00:30:53.279 --> 00:30:57.279
I recoil of this transformative use analysis that we just
541
00:30:57.640 --> 00:31:00.720
focus on so much today is you know, implicit in
542
00:31:00.720 --> 00:31:03.920
that is what is this? There's always like a second work,
543
00:31:04.319 --> 00:31:07.759
a specific second work that we're talking about whether there
544
00:31:07.839 --> 00:31:12.240
was transformation, Well, what is that in the LLM training
545
00:31:13.000 --> 00:31:18.319
module context? Again, it just doesn't fit very well. So
546
00:31:18.400 --> 00:31:21.400
again I just don't think that you know, this, the
547
00:31:21.680 --> 00:31:27.160
hanging this and this debate around a positive analysis of
548
00:31:27.200 --> 00:31:31.160
the current transformer views fair use regime makes a lot
549
00:31:31.200 --> 00:31:33.319
of sense, and it's leading us down the wrong road
550
00:31:33.359 --> 00:31:33.880
in my view.
551
00:31:34.599 --> 00:31:37.880
Yeah, just to throw it introject briefly, I think one
552
00:31:37.920 --> 00:31:40.759
area where fair use makes so clearly the most sense
553
00:31:41.559 --> 00:31:47.680
is parodies of transformations to make fun of, to critique
554
00:31:48.960 --> 00:31:52.920
works other works. Those clearly have very very strong First
555
00:31:52.960 --> 00:31:57.400
Amendment interests at state because if you don't love people
556
00:31:57.440 --> 00:32:02.160
to parody or too big political or other critique of workout.
557
00:32:01.720 --> 00:32:04.519
There, you are really limiting.
558
00:32:04.319 --> 00:32:09.160
Their First Amendment right very very severely, and that one
559
00:32:09.160 --> 00:32:13.759
could argue, I think that's something very at copying of
560
00:32:13.839 --> 00:32:17.200
code or something or thousands of lines of code, which
561
00:32:17.279 --> 00:32:21.400
is very does doesn't rise to the sort of involve
562
00:32:21.480 --> 00:32:25.880
the same First Amendment position.
563
00:32:26.079 --> 00:32:27.759
And then of course the air the question.
564
00:32:27.839 --> 00:32:30.440
I'm not going to say anything more about, you know,
565
00:32:30.480 --> 00:32:34.920
the copying used in training, but clearly, I mean, yeah,
566
00:32:35.000 --> 00:32:38.640
fair use clearly has some strong First Amendment implications that
567
00:32:38.720 --> 00:32:41.039
you should be careful how those are bounded.
568
00:32:42.599 --> 00:32:45.240
Yeah, you know, I think that there's some great arguments
569
00:32:45.240 --> 00:32:48.640
going both ways. And Jane, and I'm you know, sympathetic
570
00:32:48.680 --> 00:32:51.119
to many of the points you made, and I think
571
00:32:51.160 --> 00:32:52.599
similar arguments have been made to.
572
00:32:52.480 --> 00:32:54.920
Defend you know, the rights of these companies to train
573
00:32:55.000 --> 00:32:56.319
their models and corporate works.
574
00:32:57.200 --> 00:32:59.039
But what I find that, you know, a lot of
575
00:32:59.039 --> 00:33:03.799
these augments do rest proposition, right, the proposition that these tools,
576
00:33:03.839 --> 00:33:08.960
because that's what they are, these machine tools, do we
577
00:33:09.039 --> 00:33:12.440
perhaps tend to anthropomorphize them a little too much? You know,
578
00:33:12.960 --> 00:33:18.240
the AI is hallucinating, the AI is learning, when I
579
00:33:18.240 --> 00:33:20.920
mean humans learned too, but in a very different way, right,
580
00:33:21.000 --> 00:33:24.759
I mean, in the recent cases that I'm winding their
581
00:33:24.759 --> 00:33:26.319
way through the quotes on the subject.
582
00:33:26.720 --> 00:33:28.599
I thought it was interesting when.
583
00:33:29.920 --> 00:33:32.440
You know, a lot of the focus seems to be
584
00:33:33.200 --> 00:33:36.519
on the replication of the work. Are you creating perhaps
585
00:33:36.559 --> 00:33:39.279
a database that can then be used of future training
586
00:33:39.319 --> 00:33:44.279
and have you given proper attribution for that purpose? I
587
00:33:44.279 --> 00:33:46.759
forget the name of the case, but that was one
588
00:33:46.799 --> 00:33:49.759
of the points made by the judge recently against Anthropic.
589
00:33:51.279 --> 00:33:54.160
But I want to move on to another notion that
590
00:33:54.680 --> 00:33:57.519
frequently pops up in these cases. You know, there's this
591
00:33:57.599 --> 00:34:01.480
huge debate on what is risen transformative. In a recent
592
00:34:01.519 --> 00:34:04.200
case where the judge ruled against the rights holders suing
593
00:34:04.200 --> 00:34:10.480
the AI company H they flagged this idea that market
594
00:34:10.480 --> 00:34:13.599
dilution is going to be a very significant consideration going forward.
595
00:34:13.920 --> 00:34:17.440
To what degree is allowing this use of corporate material
596
00:34:17.880 --> 00:34:20.440
or to dilute the market for those works. This is
597
00:34:20.480 --> 00:34:23.719
kind of an analogous point to the idea of monopsny power.
598
00:34:23.760 --> 00:34:26.599
Are we suppressing the value of future output and innovation,
599
00:34:27.039 --> 00:34:30.039
which is one of the ideas being having IP law.
600
00:34:31.519 --> 00:34:33.239
So the question, you know, for anyone who wants to
601
00:34:33.559 --> 00:34:37.000
jump in on it, are we seeing any evidence one
602
00:34:37.000 --> 00:34:39.519
way or the other that these AI tools are diluting
603
00:34:40.280 --> 00:34:44.360
the market for original ideas, you know, the production of
604
00:34:44.440 --> 00:34:47.119
I think what people derisively call slop.
605
00:34:51.039 --> 00:34:53.880
I mean, I don't know that you're asking an empirical question.
606
00:34:53.960 --> 00:34:56.840
I haven't kept up with. It's an interesting question. It's
607
00:34:56.880 --> 00:34:59.000
the right question, I guess. Is how I want to
608
00:34:59.039 --> 00:35:04.519
respond that it's it's taking seriously this idea that one
609
00:35:04.559 --> 00:35:06.199
of the core, maybe not the only, but one of
610
00:35:06.280 --> 00:35:09.639
the main missions of copyright and other IP laws is
611
00:35:10.280 --> 00:35:14.000
to make sure that the incentives for future generation are
612
00:35:14.119 --> 00:35:14.800
still there.
613
00:35:15.960 --> 00:35:16.639
And I could.
614
00:35:16.480 --> 00:35:18.679
Imagine I mean, I think, I think one reason I
615
00:35:18.679 --> 00:35:21.280
don't even want to speculate about what might be happening
616
00:35:21.320 --> 00:35:23.519
empirically is that I could see it going either way.
617
00:35:23.559 --> 00:35:28.360
I could see AI bringing just a wealth of new creators,
618
00:35:28.960 --> 00:35:31.480
you know, who now have this tool that allow them
619
00:35:31.519 --> 00:35:36.199
to kind of meet the maximum potential of their creativity.
620
00:35:36.599 --> 00:35:39.039
Or I could imagine the opposite that, you know, there's
621
00:35:39.599 --> 00:35:40.280
what's the point?
622
00:35:42.960 --> 00:35:46.039
Yeah, I think that's the key way to think about
623
00:35:46.079 --> 00:35:51.480
this is again, is you know, I get I recoil
624
00:35:51.559 --> 00:35:55.400
at the debate constantly about oh, whether is AI an
625
00:35:55.440 --> 00:35:57.960
author or AI an inventor?
626
00:35:58.320 --> 00:35:58.519
You know.
627
00:36:00.159 --> 00:36:02.400
That's the wrong question. You know, the answer is no,
628
00:36:02.760 --> 00:36:04.880
They're they're never it's never going to be an author,
629
00:36:04.920 --> 00:36:06.760
it's never going to be an inventor. You know, it
630
00:36:06.880 --> 00:36:10.000
always goes back to the closest human, the human that's
631
00:36:10.239 --> 00:36:11.239
utilizing the tool.
632
00:36:11.280 --> 00:36:13.800
As you mentioned, you know, they're always going.
633
00:36:13.599 --> 00:36:16.559
To be the inventor and the author, and you know,
634
00:36:16.719 --> 00:36:19.280
and that we should encourage that because that will embrace
635
00:36:19.360 --> 00:36:21.840
the utilization of these tools in a way that's sort
636
00:36:21.840 --> 00:36:24.719
of low risk. It's not going to eliminate their livelihood,
637
00:36:25.079 --> 00:36:27.519
you know, because that's a concern, right You're going to
638
00:36:27.599 --> 00:36:29.920
have you know, AI is going to go and generate
639
00:36:30.000 --> 00:36:32.360
all these you know, all these new novels that we
640
00:36:32.400 --> 00:36:35.400
won't ever have to read, you know, uh, you know,
641
00:36:36.239 --> 00:36:38.719
pick your favorite author anymore, because we'll have all this
642
00:36:38.800 --> 00:36:41.800
stuff available to us so cheaply and and.
643
00:36:41.639 --> 00:36:42.960
You know and ubiquitously.
644
00:36:43.599 --> 00:36:47.760
You know, you know that that's again if we we
645
00:36:47.840 --> 00:36:49.880
only get there if we think about it in the
646
00:36:49.920 --> 00:36:51.760
wrong way. In my view, if we bring it back
647
00:36:51.800 --> 00:36:55.079
to there's always a human. They're the key, they're utilizing
648
00:36:55.079 --> 00:36:58.760
the tool. We'll get we'll get more and better works,
649
00:36:58.800 --> 00:37:02.000
we'll get more and better inventions. But you know, at
650
00:37:02.039 --> 00:37:05.239
the same time, we're not going to again disincentivize or
651
00:37:05.280 --> 00:37:08.400
eliminate the livelihoods of innovators that we rely on.
652
00:37:08.480 --> 00:37:11.800
America was built on innovators. We must be respectful.
653
00:37:12.079 --> 00:37:15.440
We must provide an environment where they can grow and
654
00:37:15.599 --> 00:37:18.880
prosper and you know, be fruitful and multiply that that's
655
00:37:18.920 --> 00:37:22.440
what we want. It's essential for national security, for economic security,
656
00:37:22.719 --> 00:37:23.840
or everything that we do.
657
00:37:24.079 --> 00:37:27.119
So we have to think about this stuff the right way.
658
00:37:27.159 --> 00:37:30.679
And thinking about ever that an AI can by itself
659
00:37:30.719 --> 00:37:32.960
be an author or an inventor, in my view, is
660
00:37:33.000 --> 00:37:34.880
absurd and problematic.
661
00:37:35.760 --> 00:37:39.119
Yeah, I think you know, judges, those are excellent points.
662
00:37:39.159 --> 00:37:44.039
Everyone's spaking. Judges can intelligent judges can part suit a
663
00:37:44.119 --> 00:37:47.559
copyright law. One can argue about what amendments and and
664
00:37:47.800 --> 00:37:51.880
and come to some intelligent decisions. I think thinking bartsby
665
00:37:51.920 --> 00:37:56.840
anthropic we're a judge. Also in the Northern Industrict of California,
666
00:37:56.920 --> 00:38:01.239
how using books to train Claude was exceedingly transformative in
667
00:38:01.320 --> 00:38:04.000
fair use Y.
668
00:38:04.280 --> 00:38:06.679
It had usefully.
669
00:38:07.880 --> 00:38:12.480
Scandy lawfully purchased books into a digital internal library which
670
00:38:12.559 --> 00:38:16.280
was not multiplied or redistributed, but it was It did
671
00:38:16.360 --> 00:38:23.239
inform the development of claud. However, the courts separated that
672
00:38:23.880 --> 00:38:28.880
from the acquisition and retention of books. For instance, it's
673
00:38:28.960 --> 00:38:33.320
creation of permanent library from pirated copies, and was not
674
00:38:33.480 --> 00:38:36.960
fair use. The course stata there is no AI company
675
00:38:37.000 --> 00:38:40.119
carved out from the Copyright Act and set the pirated
676
00:38:40.199 --> 00:38:44.039
coffee issue for trial on damages, So you know, you
677
00:38:44.039 --> 00:38:46.400
could read that as well. Training may be fair use
678
00:38:46.440 --> 00:38:52.000
when the model does not create output infringing copies, but
679
00:38:52.159 --> 00:38:57.920
stealing or retaining pirated source copies is a separate infringement problem.
680
00:38:57.920 --> 00:38:58.840
It is infringement.
681
00:38:59.039 --> 00:39:01.800
So I'm not saying he necessarily got it right on
682
00:39:02.159 --> 00:39:07.440
the first one. But the courts can settle, can go
683
00:39:07.519 --> 00:39:10.920
through this, and of course original eventually some of these
684
00:39:10.960 --> 00:39:13.800
issues are going to get to the Supreme Court perhaps,
685
00:39:13.840 --> 00:39:16.239
and I know Jeff from what Jeff has said, and
686
00:39:16.320 --> 00:39:17.199
I also.
687
00:39:18.400 --> 00:39:19.679
Might have all the concern you know.
688
00:39:21.119 --> 00:39:23.400
The problem is the Supreme Court has not been very
689
00:39:23.880 --> 00:39:26.559
great on property rights in general. We haven't talked about
690
00:39:26.599 --> 00:39:31.400
patent but I think the Supreme Court shockingly has ignored
691
00:39:31.400 --> 00:39:34.119
the plane statutory language of Section one on one of
692
00:39:34.159 --> 00:39:38.920
the Patent Act to find that certain certain highly novel
693
00:39:38.960 --> 00:39:43.119
inventions were not patentable, and it just made absolutely no sense.
694
00:39:43.199 --> 00:39:47.159
But there we are, so sometimes Congress does have to
695
00:39:47.199 --> 00:39:49.400
step in and hear things up.
696
00:39:50.920 --> 00:39:53.679
Thanks goes, great points, and you know, I do want
697
00:39:53.719 --> 00:39:56.360
to move on to patents in a second, but before
698
00:39:56.360 --> 00:39:58.760
I do that, I want to quickly get a response
699
00:39:58.760 --> 00:40:02.840
from this question. So the National AI Framework has recently
700
00:40:02.840 --> 00:40:05.599
been put out by the Trump administration. Obviously the goal
701
00:40:05.719 --> 00:40:09.320
is to encourage American innovation and give American companies an
702
00:40:09.360 --> 00:40:12.320
edge in the AI race. One of the things they've
703
00:40:12.719 --> 00:40:16.920
quite diplomatically tried to do is sidestep these debates about
704
00:40:16.960 --> 00:40:19.559
the fair use question and say we'll leave that answer
705
00:40:19.679 --> 00:40:20.400
to the courts.
706
00:40:20.639 --> 00:40:23.079
That plans right now to amend the corporate law.
707
00:40:23.840 --> 00:40:25.679
So my question with a panel would be, what if
708
00:40:25.679 --> 00:40:30.199
any amendments should we make to the copyright law in
709
00:40:30.320 --> 00:40:33.599
order to address these sorts of issues are being brought up,
710
00:40:33.599 --> 00:40:35.280
Because because Jeff, you know, to your point, I do
711
00:40:35.360 --> 00:40:38.079
think while the economic question can be a little bit
712
00:40:38.159 --> 00:40:40.360
murky and based on evidence that might not exist yet
713
00:40:40.679 --> 00:40:42.880
and hard to argue in each separate case, I think
714
00:40:42.920 --> 00:40:45.239
the normative or moral case is probably going to be
715
00:40:45.320 --> 00:40:47.920
what dictates any future reform in this area.
716
00:40:49.119 --> 00:40:51.880
But yeah, what amendments you know, you guys make?
717
00:40:52.280 --> 00:40:53.840
I mean, I got it, you know, I think the
718
00:40:53.880 --> 00:40:57.239
bigger question is, you know, I'm I believe, you know,
719
00:40:57.320 --> 00:41:01.559
the democratic process should should play out here. But Congress
720
00:41:02.119 --> 00:41:07.719
must set simple categorical rules, whatever they are, that is
721
00:41:07.800 --> 00:41:11.400
most important. My personal view is that it should be
722
00:41:11.480 --> 00:41:13.280
you know, the you know, in a particular case that
723
00:41:13.320 --> 00:41:16.960
we've been discussing here, that it should be defined as infringement.
724
00:41:17.239 --> 00:41:18.960
It should be infringement to.
725
00:41:19.079 --> 00:41:23.599
Utilize copyrighted works to train large language models without compensation.
726
00:41:23.840 --> 00:41:26.559
In my view, but again, I don't get to make
727
00:41:26.559 --> 00:41:27.159
that decision.
728
00:41:27.199 --> 00:41:29.960
I'm comfortable letting the But I think the more important
729
00:41:30.880 --> 00:41:34.079
from a political economy, you know, from a again, from
730
00:41:34.119 --> 00:41:37.760
a property rights perspective what have you, is we just
731
00:41:37.920 --> 00:41:44.280
need simple categorical rules because leaving it murky, leaving it
732
00:41:44.360 --> 00:41:47.199
complicated up to the courts, as we've discussed here and
733
00:41:47.480 --> 00:41:49.840
post the copyright as all you know talked about in
734
00:41:49.840 --> 00:41:51.079
the patent space, leaving it to.
735
00:41:51.039 --> 00:41:53.199
The courts is a disaster.
736
00:41:53.719 --> 00:41:57.840
And all we do is just you know, increase litigation
737
00:41:58.519 --> 00:42:02.360
and mix create confu usion, and and and and again.
738
00:42:02.519 --> 00:42:05.599
You know, again, like running home to my belief in
739
00:42:06.000 --> 00:42:09.840
the process of innovation, of the discovery process that is
740
00:42:10.000 --> 00:42:12.360
in the subjectivity. There is the core of austern economics.
741
00:42:12.400 --> 00:42:15.880
If we do that, the parties will adapt. We need
742
00:42:15.920 --> 00:42:20.719
that to believe and understand that that will happen if
743
00:42:20.760 --> 00:42:25.119
we just have the sort of institutional courage to follow
744
00:42:25.199 --> 00:42:29.320
the constitution, follow separation of powers, set the rules and
745
00:42:29.360 --> 00:42:32.239
then get out of the way and let the market drive,
746
00:42:32.800 --> 00:42:37.599
you know, uh, the adoption, the adaptation, and and.
747
00:42:37.199 --> 00:42:39.199
And the country will be better off for it.
748
00:42:39.119 --> 00:42:46.000
In my view, No, I I I really can't disigree
749
00:42:46.079 --> 00:42:50.039
with Jeff. I think particularly the pattern area, uh, that
750
00:42:50.039 --> 00:42:54.800
that's important. And you know, maybe some clear language on
751
00:42:54.800 --> 00:42:59.280
on what is fair use, let's say, conveying certain things
752
00:42:59.320 --> 00:43:03.440
that you know, very very clear categories perhaps could it
753
00:43:03.440 --> 00:43:06.920
could be helpful. I think the problem that patents is
754
00:43:06.960 --> 00:43:10.199
even most important because there's a huge amount of innovation
755
00:43:11.360 --> 00:43:17.039
driven by patents. But there are seeing medical diagnostics for example,
756
00:43:17.079 --> 00:43:19.400
or patentable in China and Germany.
757
00:43:19.880 --> 00:43:22.239
But because the court has taken had.
758
00:43:22.119 --> 00:43:28.519
Some weird interpretations about you know what our abstract ideas well.
759
00:43:28.599 --> 00:43:32.280
These are just app applying to that and a few
760
00:43:32.280 --> 00:43:37.639
other important areas you've lost. Some foreign investors have gone
761
00:43:37.639 --> 00:43:40.960
to China to do work on medical diagnostics because it's
762
00:43:41.000 --> 00:43:44.880
always too complicated and risky to try and get patents,
763
00:43:44.960 --> 00:43:49.280
defensible patents in the US. That's a terrible situation of beings.
764
00:43:49.840 --> 00:43:52.760
And unfortunately, you know, the public choice people would say,
765
00:43:52.800 --> 00:43:56.119
you've got rent seeking collisions, You've got the people who
766
00:43:56.119 --> 00:43:59.519
were using this property, we're all having to go through
767
00:43:59.559 --> 00:44:02.519
license and so on. Are going to law be against that.
768
00:44:02.719 --> 00:44:05.920
But this is very very important, I think in terms
769
00:44:05.920 --> 00:44:09.079
of international innovation and for the role of the US.
770
00:44:09.519 --> 00:44:12.400
The patent law gets straightened out by Congress.
771
00:44:14.639 --> 00:44:16.920
So you know, with regard to that, right, one of
772
00:44:16.960 --> 00:44:20.480
the debates we're having right now is around subject matter eligibility.
773
00:44:21.559 --> 00:44:24.840
Now it's rare to find a quote that has said
774
00:44:25.039 --> 00:44:28.079
blanket you know, you cannot patent this technology or that
775
00:44:28.159 --> 00:44:31.400
technology or this innovation of that innovation. Rather they will
776
00:44:31.400 --> 00:44:32.880
do it in case by case space. Is that in
777
00:44:32.920 --> 00:44:37.679
effect makes it very difficult to patent certain things. So
778
00:44:37.719 --> 00:44:40.599
for example, it is possible but difficult to patent the
779
00:44:40.639 --> 00:44:45.360
business method. It is really very difficult to patent from
780
00:44:45.400 --> 00:44:47.840
what I've read, a medical diagnostic tool.
781
00:44:49.199 --> 00:44:52.239
So the question is if we do expand this subject.
782
00:44:51.840 --> 00:44:57.920
Matter eligibility perhaps through statute, would there be any negative
783
00:44:58.000 --> 00:45:01.280
law on implications in terms of holding off innovation.
784
00:45:04.679 --> 00:45:06.719
In my view, know it would it would greatly. It
785
00:45:06.719 --> 00:45:08.320
would be you know, it would be pulled.
786
00:45:08.199 --> 00:45:12.679
Pouring gasoline on the Jackson, gasoline into the innovation engine.
787
00:45:12.719 --> 00:45:12.880
You know.
788
00:45:13.239 --> 00:45:17.280
Again the subject matter eligibility issue is the poster child
789
00:45:17.320 --> 00:45:20.559
for courts meddling where they have no business meddling, and
790
00:45:20.599 --> 00:45:24.639
the problems that is created are directly result of again violation.
791
00:45:24.760 --> 00:45:27.519
In my view of the separation of powers, Congress has
792
00:45:27.559 --> 00:45:30.000
plenary power assets the rules for patenting. They did it
793
00:45:30.039 --> 00:45:33.039
with a simple categorical rule in the text of section
794
00:45:33.119 --> 00:45:35.840
one oh one. And here come the courts say, well,
795
00:45:35.840 --> 00:45:38.880
wait a minute, you know, those four satutory categories, we
796
00:45:39.239 --> 00:45:41.400
we're gonna you know, we have to have some exceptions
797
00:45:41.400 --> 00:45:44.119
to that, and look at what that mess has created.
798
00:45:44.480 --> 00:45:47.199
And so you know, so we you know by you know,
799
00:45:47.239 --> 00:45:50.360
so we you know, you know by statute, the statutory
800
00:45:50.480 --> 00:45:52.119
change that my view will be right would be to
801
00:45:52.199 --> 00:45:56.239
eliminate abrogate all those judicial exceptions and just go back
802
00:45:56.239 --> 00:46:00.920
to the statute as written. Everybody can understand that, you know, uh,
803
00:46:01.239 --> 00:46:04.159
I've suggested you know in the meanwhile to to the
804
00:46:04.320 --> 00:46:08.639
you know, does the Federal Society Conference in November to
805
00:46:08.679 --> 00:46:11.280
the Director of the Patent Office. You know, given this
806
00:46:11.880 --> 00:46:16.119
clear violation separation powers where the judiciary is writing patent
807
00:46:16.199 --> 00:46:19.159
law in subject matter elibility context, you know, have you
808
00:46:19.280 --> 00:46:22.760
considered taking their opinions of the advisory Okay, thank you,
809
00:46:22.840 --> 00:46:24.920
We'll look at that. But you know, at the end
810
00:46:24.920 --> 00:46:26.840
of the day, we're going to interpret the you know,
811
00:46:26.880 --> 00:46:30.159
we can interpret. We're article to agency. We can you know,
812
00:46:30.480 --> 00:46:34.760
we can interpret that statute, and we appreciate your opinion,
813
00:46:34.760 --> 00:46:36.920
but we'll make the cause of whether to cancel the
814
00:46:36.920 --> 00:46:39.159
pattern or not. You know, this is the kind of
815
00:46:39.159 --> 00:46:41.760
thing that we you know again that you know, we
816
00:46:41.920 --> 00:46:45.159
you know, it just illustrates again, you know, the problems
817
00:46:45.159 --> 00:46:49.760
that are created when courts go outside their lane and
818
00:46:50.400 --> 00:46:52.840
start to again violate separation powers.
819
00:46:52.880 --> 00:46:56.199
In my view, can I maybe our way in to,
820
00:46:56.639 --> 00:46:59.639
you know, try to find a little bit of separation
821
00:46:59.760 --> 00:47:04.440
or diversy between us. I so I'm highly sympathetic to
822
00:47:04.519 --> 00:47:06.559
some of the examples that have come out, like the
823
00:47:06.920 --> 00:47:13.559
medical diagnosis devices for example, and I mentioned new news
824
00:47:13.599 --> 00:47:16.920
gathering earlier, like there are certain types of even raw
825
00:47:16.960 --> 00:47:20.159
information that today isn't even copyrightable, that I think we
826
00:47:20.239 --> 00:47:24.920
have a serious incentive problem. But my concern is that,
827
00:47:25.880 --> 00:47:27.840
you know, what those things have in common is that
828
00:47:27.960 --> 00:47:33.039
it's surfacing information that's hard to extract otherwise it takes
829
00:47:33.079 --> 00:47:38.199
effort or ingenuity, and it's highly useful. And so then
830
00:47:38.239 --> 00:47:40.440
the you know, then then the problem of free writing
831
00:47:40.639 --> 00:47:43.639
is right there, and and so I think we all
832
00:47:43.639 --> 00:47:46.159
are interested in solving that. But on the other side,
833
00:47:46.159 --> 00:47:48.639
when I think about who would be lobbying Congress for
834
00:47:48.800 --> 00:47:53.079
these changes, it's you know, it's going to be you know,
835
00:47:53.119 --> 00:47:55.599
if I think about what the RIAA does, it's like,
836
00:47:56.199 --> 00:47:59.599
do we really need more incentive for people to become
837
00:47:59.679 --> 00:48:01.519
select bridies and recording artists?
838
00:48:01.599 --> 00:48:02.880
I actually don't think we do.
839
00:48:03.000 --> 00:48:05.679
And perhaps, you know, maybe Jeff and I can compromise
840
00:48:05.719 --> 00:48:09.159
where maybe if the terms of copyright are changed or
841
00:48:09.440 --> 00:48:13.679
you know, modified to uh TO kind of fit in
842
00:48:13.880 --> 00:48:18.880
categorical terms, how how much other sort of psychic rewards
843
00:48:18.920 --> 00:48:22.159
or how much other incentive there is for people to produce?
844
00:48:22.559 --> 00:48:24.199
You know, maybe I maybe.
845
00:48:24.039 --> 00:48:25.559
It would make sense, I think to.
846
00:48:27.159 --> 00:48:31.360
Uh TO to expand uh to expand rights. But but
847
00:48:31.400 --> 00:48:33.800
I do worry a bit, being in political economy, about
848
00:48:33.800 --> 00:48:37.920
who exactly is going to be there ready to get
849
00:48:37.960 --> 00:48:40.239
their their sort of favored terms.
850
00:48:42.880 --> 00:48:46.800
Yeah, that's interesting, right, because I mean, leaving aside, you know,
851
00:48:46.840 --> 00:48:50.639
I think the moral question of compensating people for their
852
00:48:50.719 --> 00:48:54.920
effort and for their creativity. To me, it seems like
853
00:48:54.960 --> 00:48:58.800
if an AI tool can ingest the sum total of
854
00:48:58.840 --> 00:49:03.000
your works and produce something generic that replicates them in
855
00:49:03.039 --> 00:49:06.239
a way that's as good as something you yourself will continue
856
00:49:06.239 --> 00:49:08.239
to put out, that's probably a sign that you, as
857
00:49:08.239 --> 00:49:11.679
an artists need to do something different, something more, or
858
00:49:12.440 --> 00:49:15.119
you know, perhaps your art or your style happens to
859
00:49:15.119 --> 00:49:18.920
be a particularly generic one. And that being said, for
860
00:49:18.960 --> 00:49:20.559
all the talk about eyeslope, I do think there's a
861
00:49:20.599 --> 00:49:21.760
lot of human slope out there.
862
00:49:21.840 --> 00:49:23.800
You know, mumble prap is an example.
863
00:49:23.840 --> 00:49:25.519
But I mean that's just you know, I'm biased on
864
00:49:25.639 --> 00:49:29.519
this topic, so you know, I do think that there
865
00:49:29.599 --> 00:49:32.639
is an argument, least in the economics aside, leaving aside
866
00:49:32.639 --> 00:49:35.119
of the question of compensation for the use of IP,
867
00:49:36.639 --> 00:49:39.880
that these tools will little long run perhaps encourage more
868
00:49:39.920 --> 00:49:44.199
creativity because you now know what the generic version of
869
00:49:44.199 --> 00:49:46.880
something sounds like, and that makes it easier to get
870
00:49:46.880 --> 00:49:48.440
to Okay, what's innovative, what's new?
871
00:49:48.719 --> 00:49:52.000
What's next? What can I add in total? Now exists?
872
00:49:53.639 --> 00:49:57.199
All right, So we move on to audience questions. There
873
00:49:57.320 --> 00:50:01.039
some interesting questions here. So the opinion.
874
00:50:01.400 --> 00:50:04.880
The opinion sauto move the court away from a purely
875
00:50:04.880 --> 00:50:05.760
commercial focus.
876
00:50:06.519 --> 00:50:07.679
Uh, this is a question.
877
00:50:07.519 --> 00:50:10.679
From doctor Jack David Gulub move the quurte away from
878
00:50:10.679 --> 00:50:13.679
a purely commercial focus towards determining if a new work
879
00:50:14.000 --> 00:50:16.679
supersedes the original creates something of value for the public.
880
00:50:17.599 --> 00:50:20.719
Is this not analogous to the patent law embodying the
881
00:50:20.760 --> 00:50:22.480
novelty doctrine.
882
00:50:29.440 --> 00:50:31.360
We've really jumped in of anyone as well?
883
00:50:33.480 --> 00:50:39.559
Really analogous to the pad law. I guess I'm just
884
00:50:39.599 --> 00:50:43.519
not exactly seeing what the specific analogy here. Maybe I
885
00:50:43.599 --> 00:50:45.320
missed miss that, but.
886
00:50:47.599 --> 00:50:57.920
I mean, okay, it's creating a new market basically because
887
00:50:57.960 --> 00:51:01.519
it's doing something that goes beyond maybe embedding, but goes.
888
00:51:01.320 --> 00:51:02.960
Beyond Is that I don't know.
889
00:51:04.559 --> 00:51:07.440
I don't know enough about patent laws, So trying to
890
00:51:07.480 --> 00:51:11.480
help translate, was the question about.
891
00:51:11.400 --> 00:51:14.119
Creating a new work that is, you know, sort of
892
00:51:16.079 --> 00:51:18.719
sufficiently different than the previous work? Kind of like, is
893
00:51:18.760 --> 00:51:20.440
it the analogy to obviousness?
894
00:51:20.480 --> 00:51:20.679
Is that?
895
00:51:20.760 --> 00:51:25.320
Is that what I'm understanding. I mean, you have the
896
00:51:25.440 --> 00:51:33.159
substantially similar doctrine in copyright right before you can before
897
00:51:33.199 --> 00:51:35.119
you can infringe.
898
00:51:35.440 --> 00:51:39.280
I guess that's something to think about it in that context.
899
00:51:40.400 --> 00:51:43.639
But you know, but you know, I don't know.
900
00:51:43.760 --> 00:51:47.880
In in copyright right, it's it's about the expression, So
901
00:51:48.840 --> 00:51:51.199
you know, express the idea in a different way.
902
00:51:51.480 --> 00:51:53.760
That's a different work, you know, and it's intimes.
903
00:51:53.840 --> 00:51:56.360
Do you think derivative work the concept of derivative works
904
00:51:56.440 --> 00:51:59.719
might be too broad if we compare how patent law
905
00:52:00.440 --> 00:52:03.199
maybe that's kind of the essence of the question. If
906
00:52:03.199 --> 00:52:06.559
we compare how patent law deals with new kind of
907
00:52:06.599 --> 00:52:09.840
add ons, does that imply that maybe derivative works the
908
00:52:09.840 --> 00:52:11.360
scope of derivative works is too broad?
909
00:52:12.480 --> 00:52:14.760
Well, yeah, I mean clearly is are right you have
910
00:52:14.840 --> 00:52:16.679
a right to as a cover in order to can
911
00:52:16.800 --> 00:52:22.599
come up with derivative works in patents there it's somewhat different.
912
00:52:22.639 --> 00:52:25.320
People can use the big thing of a panel, they
913
00:52:25.360 --> 00:52:29.440
can use the information, the public can use the information disclosed.
914
00:52:29.480 --> 00:52:32.400
That is one of the arguments as to why patents
915
00:52:32.480 --> 00:52:36.320
maybe social superior, socially superior to trade secrets. They allow
916
00:52:36.400 --> 00:52:41.039
that information out there. But you just but and people
917
00:52:41.039 --> 00:52:43.320
are allowed to use that information to come out and say, well,
918
00:52:43.360 --> 00:52:45.440
their own patents, which are.
919
00:52:45.559 --> 00:52:46.760
But they have to be novel.
920
00:52:48.079 --> 00:52:51.079
They have to be you know, not based on substantially
921
00:52:51.119 --> 00:52:52.159
based on priority.
922
00:52:52.239 --> 00:52:54.880
You have to go beyond that. So you have to
923
00:52:54.880 --> 00:52:56.280
show your own form.
924
00:52:56.199 --> 00:53:00.719
Of novelty and maybe able to get your own patterns.
925
00:53:00.760 --> 00:53:02.159
Is a pattern of holder.
926
00:53:01.920 --> 00:53:04.679
May not an irrivative work, but they have a follow
927
00:53:04.719 --> 00:53:08.679
on patent that that has to meet these conditions of
928
00:53:08.760 --> 00:53:12.280
patent of building. Say the way you change a DOUG
929
00:53:12.360 --> 00:53:16.000
drug may be chemical chemically injected. If it's obvious you
930
00:53:16.039 --> 00:53:18.159
can't do that, but you may be able to come
931
00:53:18.239 --> 00:53:21.559
up with some changes that allow you for new patents.
932
00:53:21.599 --> 00:53:23.760
So again there yet there's sort of it in an
933
00:53:23.760 --> 00:53:26.920
analogy based not it doesn't really work to quote the
934
00:53:26.920 --> 00:53:29.000
same way as it works because of the.
935
00:53:30.559 --> 00:53:34.880
Two statutory schemes are fairly different. Cool.
936
00:53:36.239 --> 00:53:39.199
We have a question on corporate terms. Any opinion on
937
00:53:39.239 --> 00:53:43.159
whether corporate terms are too long? Isn't it limited a
938
00:53:43.280 --> 00:53:46.400
time properly determined as a portion of human life? Otherwise,
939
00:53:46.400 --> 00:53:48.519
if there is no quid pro quo, a busn't the
940
00:53:48.599 --> 00:53:51.000
twenties today will never see fair use of corporated materials
941
00:53:51.039 --> 00:53:53.960
they spend money on in theory, but in theory, by
942
00:53:54.000 --> 00:53:56.440
the time someone in the twenties today should see fair
943
00:53:56.519 --> 00:53:58.960
use of today's data by time they are fifty.
944
00:54:00.360 --> 00:54:02.119
I want to I want to explain the question a
945
00:54:02.159 --> 00:54:02.599
little bit.
946
00:54:04.239 --> 00:54:09.159
Does thoughts on the current duration for corporate terms as
947
00:54:09.159 --> 00:54:09.679
well as.
948
00:54:09.599 --> 00:54:16.039
Patent Well, again, you know, they're they're they're currently simple
949
00:54:16.039 --> 00:54:18.760
and categorical. We know what they are, you know, and
950
00:54:18.800 --> 00:54:21.199
they've been a little bit muddled in the patent space.
951
00:54:21.559 --> 00:54:25.000
It's you know, things you know, with extensions and adjustments
952
00:54:25.039 --> 00:54:29.440
and obviously the type of double patenting which again mostly
953
00:54:30.039 --> 00:54:33.159
judge made meddling that have caused those problems.
954
00:54:34.320 --> 00:54:37.480
But again, you know, I get back to that.
955
00:54:37.559 --> 00:54:40.559
You know, as long as they're simple and categorical, Uh,
956
00:54:40.599 --> 00:54:43.039
you know, I can live with whatever they are. The
957
00:54:43.079 --> 00:54:46.599
market process, the market will adapt based on whatever we
958
00:54:46.760 --> 00:54:50.039
decide as a you know, as a body politic that
959
00:54:50.079 --> 00:54:50.679
they should be.
960
00:54:50.760 --> 00:54:51.840
But that that's the key.
961
00:54:51.960 --> 00:54:55.039
Don't complicate them, don't have a bunch of exceptions, don't
962
00:54:55.039 --> 00:54:59.000
have you know, uh, different different terms for different kinds
963
00:54:59.039 --> 00:55:02.159
of works, you know, or but again again or if
964
00:55:02.159 --> 00:55:04.400
you do again, have to be very clear and very straightforward.
965
00:55:05.079 --> 00:55:08.800
Yeah, the Supreme Court did address the issues there'd be
966
00:55:08.840 --> 00:55:10.480
something too long of a term.
967
00:55:10.719 --> 00:55:13.000
It was an infamous Mickey Mouse case.
968
00:55:13.519 --> 00:55:16.320
I think Disney had lobby to get an extension where
969
00:55:16.360 --> 00:55:19.920
with seventy years seventy five years beyond the term of death,
970
00:55:20.039 --> 00:55:23.599
because they wanted to preserve copyrights for some early Mickey
971
00:55:23.599 --> 00:55:25.719
Mouse cartoons and copyrighting.
972
00:55:25.800 --> 00:55:28.519
Yeah, they went around the market with that one. They
973
00:55:28.599 --> 00:55:30.280
went around the market.
974
00:55:29.679 --> 00:55:31.320
Because it was a matter of a blobbing.
975
00:55:31.320 --> 00:55:38.280
But the court basically didn't say did not say, oh.
976
00:55:36.519 --> 00:55:39.920
We can tell that this is it's no longer limited.
977
00:55:39.920 --> 00:55:42.639
Now if you had a thousand year term or ten
978
00:55:42.679 --> 00:55:46.760
thousand year tear, that might be. But they correctly if
979
00:55:46.760 --> 00:55:49.719
i'm long tam, was it seventy five or was that
980
00:55:49.719 --> 00:55:54.480
the magic amount? Of court refused to opine and suggest
981
00:55:54.519 --> 00:55:56.239
that that was not limited.
982
00:55:57.679 --> 00:56:04.920
Right, So there's a question about how much of a
983
00:56:05.000 --> 00:56:09.360
human contribution to create a work should enable the use
984
00:56:09.400 --> 00:56:11.800
of an AI tool to be named as an author
985
00:56:11.840 --> 00:56:14.760
of something original. And on the patent side, will the
986
00:56:14.840 --> 00:56:19.519
test be quote substantial contribution under the new case or
987
00:56:19.599 --> 00:56:22.400
a conception of all claimed limitations under the same case.
988
00:56:23.039 --> 00:56:25.880
It appears in many cases no human will satisfy the criteria,
989
00:56:26.400 --> 00:56:28.599
and thus the property right will be invalidated as not
990
00:56:28.679 --> 00:56:29.800
being the product of a human.
991
00:56:31.320 --> 00:56:36.079
That's I mean, I remarked on this during the PAN
992
00:56:36.360 --> 00:56:38.519
I think this is the wrong way to think about it.
993
00:56:39.039 --> 00:56:43.159
You know that we need to sort of define state
994
00:56:43.480 --> 00:56:47.000
unequivocally that you know, AI.
995
00:56:46.800 --> 00:56:49.880
Tools cannot be on their own authors and inventors. They
996
00:56:49.920 --> 00:56:50.960
are always tools.
997
00:56:51.000 --> 00:56:53.480
There's always a human and you just have to find
998
00:56:53.559 --> 00:56:57.239
that human and make them him or her the author
999
00:56:57.519 --> 00:57:02.440
or the inventor. And then this debate goes away and
1000
00:57:02.880 --> 00:57:09.440
we you know, energize innovation because we are allowing the
1001
00:57:09.840 --> 00:57:13.079
tools that we have available us to be used in
1002
00:57:13.159 --> 00:57:16.880
their most effective and their maximized way, just like you know,
1003
00:57:16.880 --> 00:57:20.199
when we invented you know, different laboratory equipment or other
1004
00:57:20.239 --> 00:57:23.679
things to help the discovery of new pharmaceuticals, or you know,
1005
00:57:23.719 --> 00:57:26.760
we invented the personal computer to help authors, you know,
1006
00:57:27.320 --> 00:57:29.679
put their put their work on paper. You know, these
1007
00:57:29.800 --> 00:57:33.199
kinds of things, you know, are again their tools to
1008
00:57:33.440 --> 00:57:37.760
drive to assist the human innovator. And that's the way
1009
00:57:37.800 --> 00:57:39.639
it needs to be thought about. We should not be
1010
00:57:39.719 --> 00:57:42.079
debating and say, you know, if you just say no
1011
00:57:42.199 --> 00:57:44.239
that this is not going to happen. We're not going
1012
00:57:44.280 --> 00:57:46.119
to do this. This is not possible. It's now we
1013
00:57:46.159 --> 00:57:50.199
think of it. It's always the human and the AI
1014
00:57:50.679 --> 00:57:51.280
is the tool.
1015
00:57:52.199 --> 00:57:52.480
Jeff.
1016
00:57:52.480 --> 00:57:55.800
Do you think though, that there might be some instances
1017
00:57:55.840 --> 00:58:01.840
where the human contributions are so difi to diffuse or
1018
00:58:01.880 --> 00:58:05.239
indirect that we should just treat it, treat the output
1019
00:58:05.480 --> 00:58:07.639
as immediately within the public domain.
1020
00:58:07.719 --> 00:58:10.360
So I'm thinking about there are some like there.
1021
00:58:10.199 --> 00:58:14.199
Are some uh I don't know, like hacker groups that
1022
00:58:14.320 --> 00:58:16.920
right now that are using AI to just generate a
1023
00:58:16.960 --> 00:58:22.159
whole bunch of you know, patent ideas or copyrightable you
1024
00:58:22.199 --> 00:58:25.000
know expression and and they're not you know, using them,
1025
00:58:25.039 --> 00:58:27.719
They just are are creating these giant bodies.
1026
00:58:28.960 --> 00:58:29.360
Uh.
1027
00:58:30.639 --> 00:58:34.199
So you know, if if something's automated enough, do you
1028
00:58:34.280 --> 00:58:37.559
think that maybe we should recognize no author and no inventor.
1029
00:58:39.760 --> 00:58:42.519
Well, you know, yeah, there's always you know, an issue
1030
00:58:42.519 --> 00:58:45.199
of nefarious actors getting in and we you know, we
1031
00:58:45.239 --> 00:58:47.480
do worry about that in the patent space, about you know,
1032
00:58:47.920 --> 00:58:50.559
you know, generating a bunch of prior arts. You know,
1033
00:58:50.719 --> 00:58:53.960
I could undermine the ability to you know, to get inventions,
1034
00:58:54.280 --> 00:58:58.079
you know, uh patented. But but I think Uh, there's
1035
00:58:58.119 --> 00:59:01.400
a better way to deal with that, you know, aside
1036
00:59:01.440 --> 00:59:04.199
from you know, again claiming that the AI is the
1037
00:59:04.199 --> 00:59:06.559
author and then it's not eligible because I think we're
1038
00:59:06.599 --> 00:59:10.000
seeing where that leads to and it's just it's too messy.
1039
00:59:10.199 --> 00:59:13.960
There's better ways to handle uh, you know that that
1040
00:59:14.039 --> 00:59:17.679
kind of you know, sort of computer generated arts of
1041
00:59:17.679 --> 00:59:18.960
the examination process.
1042
00:59:19.519 --> 00:59:22.639
Uh, And that's that's kind of how I would address that.
1043
00:59:24.400 --> 00:59:26.239
All right, we're almost right at time.
1044
00:59:26.920 --> 00:59:28.599
Thank you all so much for what's been a really
1045
00:59:28.639 --> 00:59:32.039
read and insightful discussion, and yeah, looking forward to the
1046
00:59:32.039 --> 00:59:32.440
next one.
1047
00:59:34.239 --> 00:59:34.800
Wonderful.
1048
00:59:35.000 --> 00:59:38.119
Well, thank you all so much for presenting on today's
1049
00:59:38.119 --> 00:59:41.320
program for this suing your expertise and opinions on this
1050
00:59:41.360 --> 00:59:44.519
really important topic, and thank you to Satya for moderating.
1051
00:59:45.360 --> 00:59:47.679
If our audience is interested in learning more about all
1052
00:59:47.719 --> 00:59:50.320
of our programming here at the Federalist Society, please visit
1053
00:59:50.320 --> 00:59:53.239
our website that is f e D Associate dot org,
1054
00:59:53.559 --> 00:59:57.000
fed SALC fedsock dot org, or feel free to send
1055
00:59:57.079 --> 00:59:59.239
us your feedback. We always love to hear it at
1056
00:59:59.239 --> 01:00:01.840
fedsock at fedsoc dot org.
1057
01:00:02.119 --> 01:00:03.320
And with that we are JERD.
1058
01:00:03.559 --> 01:00:05.800
Thank you so much, Thank you for listening.
1059
01:00:05.840 --> 01:00:08.559
To this episode of fedsoc Forums, a podcast of the
1060
01:00:08.599 --> 01:00:12.599
Federal Societies Practice Groups. For more information about the Federal Society,
1061
01:00:12.800 --> 01:00:15.679
the practice groups, and to become a Federal Society member,
1062
01:00:15.719 --> 01:00:18.440
please visit our website at fedsoc dot org.