Creativity in the Age of AI: Generative AI Issues in Art Copyright & Open Source

Summary

The panel discussion delves into the intersection of open source, copyright, and generative AI, highlighting the challenges and opportunities these technologies pose. Scott Draves, an AI artist and engineer, shares his experiences with open source generative art and its impact on the art world. Pam Samuelson, a law professor, addresses the legal complexities surrounding copyright infringement and the use of copyrighted works as training data for AI. The conversation explores the potential of generative AI to transform personal expression, education, and creativity, while also considering the need for trust, accountability, and the role of open source in democratizing AI technology.

Takeaways

• 🎨 The panel discussion focused on the intersection of Open Source, copyright, and generative AI, highlighting the complexities and ongoing legal debates surrounding these issues.
• 🤖 Scott Draves, an AI artist and engineer, discussed the history of generative AI and its evolution from his personal experiences, emphasizing the importance of open source in enabling artistic and technological innovation.
• 🌐 Draves mentioned his early work, including the 'flame' algorithm, which became a foundational piece in generative art and was an example of the collaborative spirit of the open source community.
• 🚀 The development of AI technology has seen a significant leap in recent years, with implications for personalized education, new forms of personal expression, and interactive media.
• 🧠 Draves argued for the necessity of open source models in AI, stating that having transparent and accessible code, data, and models fosters trust, independent research, and equitable distribution of AI benefits.
• 📚 The talk by Pam Samuelson, a law professor, addressed the copyright challenges posed by generative AI, noting that there are active lawsuits and global conversations about the governance of AI.
• ⚖️ Samuelson discussed the five exclusive rights granted by copyright and the limitations and exceptions to these rights, such as fair use, which is a defense often raised in cases involving AI and copyright.
• 🔍 The issue of whether AI-generated outputs infringe on derivative works was highlighted, with ongoing lawsuits challenging the boundaries of copyright law in the context of AI.
• 🎵 The discussion touched on the potential legal ramifications of AI-generated music covers, noting the complexities of copyright infringement versus creative expression and the right of publicity.
• 🌐 The panelists acknowledged that the legal landscape for AI and copyright is still developing, with technology often outpacing the law and requiring adaptation and new interpretations of existing legal frameworks.
• 📖 The Warhol vs. Goldsmith case was mentioned as an example of the ongoing legal discourse around fair use and transformative works, with implications for the use of AI in creating new works based on existing content.

Q & A

What is the main focus of the panel discussion in the transcript?

The main focus of the panel discussion is on Open Source and copyright issues related to generative AI, particularly in the context of art and its legal implications.

Who is moderating the panel and what are their background and roles?

The panel is moderated by Professor Russ Altman, who is the Kenneth Fong professor and professor of bioengineering, genetics, medicine, and biomedical data science. He is also an associate director of Stanford Hai.

What is the significance of having a bioengineer faculty moderate the panel on generative AI and copyright?

The significance lies in the interdisciplinary nature of the topic, as generative AI intersects with both technology and legal domains, and a bioengineer brings a unique perspective that can enhance the understanding of these complex issues.

What is the background of Scott Draves, the first speaker introduced in the transcript?

Scott Draves is an AI artist and engineer known for his work in open-source and generative art since the 80s and 90s. He is recognized for creating the 'flame' algorithm, which became a significant visual style and influenced a whole genre of art.

What is the Electric Sheep project mentioned by Scott Draves?

Electric Sheep is an infinite, evolving AI screensaver created by Scott Draves that uses feedback from its audience to learn and generate animations, representing an early example of collective intelligence in AI.

What are the four key tenets of open source that Scott Draves discusses in relation to AI models?

The four key tenets are: the code for training the models should be open, the inference engine code should be open, the training data should be open, and the model weights should be open.

What is the main concern regarding the use of copyrighted works as training data for AI models?

The main concern is whether ingesting copies of copyrighted works as training data constitutes an infringement of copyrights in those works, as most content on the open web is protected by copyright law.

What is the significance of the fair use defense in the context of AI-generated outputs?

The fair use defense is significant because it allows for the non-infringing use of copyrighted works under certain conditions, such as for educational purposes, criticism, commentary, news reporting, and research. It is often invoked when discussing the ingestion of copyrighted works as training data for AI models.

What is the role of the United States Copyright Office in addressing generative AI issues?

The United States Copyright Office is holding listening sessions to gather stakeholders' interests and concerns about generative AI and copyright. It will publish a report with recommendations, potentially by the end of the year, to guide its approach to these issues.

How does the concept of 'forking' and fine-tuning of AI models contribute to the diversity of truth and beauty in AI-generated content?

Forking and fine-tuning allow individuals or groups to take a foundation model, add more data, and change its character, which can lead to different notions of truth and beauty. This enables a variety of perspectives and cultural expressions in AI-generated content, rather than a single, uniform output from a single company's model.

What is the potential impact of generative AI on personal expression and creativity?

Generative AI can lead to new forms of personal expression, such as easy-to-use Photoshop alternatives that respond to user commands, interactive fiction, and personalized content creation. It has the potential to democratize creativity and enable individuals to express themselves in novel ways.

Outlines

00:00

🎨 Introduction to Generative AI and Open Source

The speaker begins by expressing gratitude for being part of the panel and introduces Scott Draves, an AI artist and engineer, and Pam Samuelson, a law professor and co-director of the Center for Law and Technology. The discussion focuses on generative AI, its history, and its implications on art and copyright, particularly in the context of open source. Scott shares his journey in creating open source art since the 80s and 90s, highlighting the transformative impact of his work on the art world and the concept of collaborative creation.

05:03

💡 The Evolution of Generative AI and its Impact

The speaker delves into the evolution of generative AI, from his early attempts at creating algorithms that generate images to the current state of the technology. He discusses the potential applications of AI, such as personalized education and assistance, and emphasizes the importance of open source in democratizing AI. The speaker argues that open source enables customization, trust, and widespread adoption of AI, while also addressing the challenges of trust and accountability in proprietary models.

10:04

📚 Copyright Challenges in the Age of Generative AI

The speaker shifts focus to the legal and policy challenges surrounding generative AI, particularly copyright issues. He outlines the current lawsuits against generative AI, the concerns of policymakers, and the ongoing discussions about the role of copyright in the digital age. The speaker explains the basics of copyright law and its limitations, and discusses the fair use doctrine as a potential defense for using copyrighted works as training data for AI.

15:05

🤖 The Role of Generative AI in Creative Reuse

The speaker explores the role of generative AI in enabling creative reuse of copyrighted works. He argues that while some view AI-generated outputs as infringing derivative works, others see it as a tool that promotes the progress of science and culture. The speaker also touches on the complexities of defining 'original expression' in the context of AI and the potential for AI to create new forms of creative expression that do not directly copy existing works.

20:06

🔍 Navigating the Legal Landscape of Generative AI

The speaker discusses the ongoing legal disputes surrounding generative AI, particularly in the music industry. He highlights the challenges of regulating AI-generated music covers and the potential implications of right of publicity and false representation laws. The speaker emphasizes the need for a nuanced approach to copyright and intellectual property in the context of AI, recognizing the balance between protecting creators' rights and fostering innovation.

25:08

🌐 The Future of Generative AI and Copyright

The speaker concludes by speculating on the future of generative AI and copyright. He addresses the potential impact of AI on the viability of fair use claims and the need for a balanced approach to legal frameworks. The speaker suggests that while existing laws may need to evolve to accommodate AI, the focus should be on fostering innovation and creativity while respecting the rights of creators.

Keywords

💡Generative AI

Generative AI refers to artificial intelligence systems that are capable of creating new content, such as images, music, or text, without direct human input. In the video, the speaker discusses the history and development of generative AI, its impact on art and copyright, and the potential for open-source models to democratize access to this technology.

💡Open Source

Open source refers to a philosophy and practice of allowing users to access, use, modify, and distribute software freely. The speaker advocates for open-source principles in AI development, emphasizing the importance of open code, open data, and the ability for users to fine-tune models to their needs.

💡Copyright

Copyright is a legal right that grants creators exclusive control over their original works. The discussion in the video revolves around the challenges that generative AI poses to traditional copyright concepts, including the use of copyrighted works as training data and the potential for AI-generated outputs to infringe on derivative works.

💡Fair Use

Fair use is a legal doctrine that allows limited use of copyrighted material without permission from the rights holder, typically for purposes such as criticism, commentary, news reporting, teaching, scholarship, or research. The video explores the application of fair use to the training of generative AI systems and the use of copyrighted works as data.

💡Creative Commons

Creative Commons is a non-profit organization that helps creators share their work under certain standardized licenses that allow others to use the work under defined conditions, such as attribution, non-commercial use, or no derivatives. The video discusses the position of Creative Commons on the use of their licensed works in AI training data.

💡Data Licensing

Data licensing refers to the legal agreements that define how data can be used, shared, and monetized. The video touches on the increasing recognition of data's value and the creation of licensing agreements for using data in AI training, which may impact the application of fair use.

💡Right of Publicity

The right of publicity is a legal right that allows individuals to control the commercial use of their name, image, likeness, or other recognizable aspects of their identity. The video highlights the potential for this right to come into play with generative AI, especially in cases where AI-generated content might use a celebrity's voice or likeness without permission.

💡Foundation Models

Foundation models are large-scale AI models that are pre-trained on vast amounts of data and can be fine-tuned for specific tasks or applications. The video emphasizes the cost-effectiveness of fine-tuning these models and the potential for diverse customization by different users or groups.

💡Trust and Accountability

Trust and accountability pertain to the reliability and responsibility associated with AI systems, particularly in terms of their transparency, the ability to audit their operations, and the consequences of their actions. The video discusses the importance of open-source AI in fostering trust and accountability by allowing independent research and third-party audits.

💡Liberal Licensing

Liberal licensing refers to the practice of granting permissions for the use of a work with few restrictions, often allowing for commercial use and modification. The video highlights how liberal licensing can encourage the spread of AI benefits and motivate investment in open-source projects.

Highlights

Discussion on Open Source and copyright in relation to generative AI.

Introduction of panel moderator, Professor Russ Altman, and his diverse academic and professional roles.

Scott Draves, AI artist and engineer, shares his journey in open source and generative art since the 80s and 90s.

The significance of 'flame number 149' as a pivotal piece in the history of open source artwork.

Generative AI's role in enabling artists and fostering meta interactivity through programming and internet collaboration.

The Electric Sheep project as an example of collective intelligence created 24 years ago and its ongoing relevance.

Scott Draves' perspective on the rapid advancements in AI's understanding of human language and image generation.

The potential implications of generative AI, including personalized education and new forms of personal expression.

The importance of open source in AI development, including transparency, trust, and the ability to fine-tune models.

Pam Samuelson discussing copyright challenges posed by generative AI and the ongoing global conversation.

Three active lawsuits challenging generative AI on copyright and related grounds, and the U.S. Copyright Office's efforts to address these issues.

The critical questions around ingesting copyrighted works as training data, AI-generated outputs as derivative works, and removal or alteration of copyright management information.

The role of fair use in the context of using copyrighted works for training data, and its legal precedents.

The distinction between AI's interest in facts versus original expression in a work, and how this impacts copyright considerations.

The potential market competition between AI-generated images and those created by human artists.

The impact of generative AI on the rights of publicity and the potential legal issues arising from the use of samples of voices or likenesses.

The Warhol versus Goldsmith case and its limited impact on the broader discussion of AI and copyright.

Assessment of the viability of fair use claims in the context of AI, given the increasing commercial value of data.

The importance of forking and fine-tuning in open source models to allow for diverse interpretations of truth and beauty.

Transcripts

that was great and we're going to

continue on in this next panel talking

more about Open Source and copyright so

I want to first introduce our panel

moderator uh Professor Russ Altman Russ

is the Kenneth Fong professor and

professor of bioengineering of genetics

of Medicine of biomedical data science

and by courtesy of computer science he's

also one of the associate directors of

Stanford Hai and uh when I met Russ I

felt my long-lost brother so thank you

Russ

thank you thank you James uh this

session is called generative AI issues

in art copyright and open source so

what's more natural than having a

bioengineer faculty moderate it

so I'm really happy to uh very uh

briefly introduce Scott Draves who will

speak first he's an AI artist and

engineer and Pam Samuelson will speak

after Scott she's a professor of law at

Berkeley law school and co-director of

Center for Law of the Center for Law and

technology so um with those brief

introductions I'll just throw it to

Scott

thank you uh and thank you thanks for uh

all your attention this morning it's uh

it's an honor to be here so a lot of

people have a take on on generative Ai

and uh before I give you yet another one

um just uh where do you know where does

my where does mine come from and where

do I come from and uh I was trying to

think of you know what I can best

contribute here uh today and I thought

one thing is just you know where where

is where does generative AI come from

what's the the history here

and so this image here's an example uh

you know I've been doing this open

source and generative uh art since the

80s and 90s and uh this is uh this image

is uh flame number 149 uh from 1994

where it got a an award at the pre-arse

Electronica and is really what made me

realize I you know I was an artist

before uh that was

and um

this uh was perhaps the you know the

first open source artwork and it's was

really very much it wasn't just an image

it was it was code that allowed other

artists to create their own images

and this artwork was very much about

enabling other artists and um the sort

of meta interactivity of uh you know

programming exchanging code over the

internet and uh you know building things

together

and so this flame algorithm uh as as it

became known created a whole genre and

visual style and eventually I couldn't

even go into a bookstore or like a look

at a magazine rack without without

seeing some version of it and of course

so

uh

there we go here we go

uh so uh here's an album cover uh that

used it and of course these were all

images uh I didn't get any sort of

credit credit for this my name isn't

isn't here it was made by another artist

just using the algorithm

and um you know really what I learned

from this was uh you know I put this

code out there with a plan

um

but what I learned was that uh sort of

giving up control uh the real power

isn't sort of getting what you want

getting other artists to use it or to

contribute to give me code back to uh

but really getting things that I didn't

know I wanted or or I didn't want and

enabling other people to do things uh

that were really unexpected

and so just uh by in 1999

uh the Electric Sheep was the I created

this infinite evolving AI screensaver

you know which used feedback from its

audience and so it learned from

everybody who was watching it and it

tried to satisfy that that human desire

and it generated animations

excuse me and um this was a collective

intelligence created uh 20 24 years ago

and actually still running and so this

this image

sorry uh here we go

and so this image is actually a still

frame from a Super Bowl commercial that

IBM and uh H R Block Rand so

it definitely went places that that I

didn't expect and and didn't plan

so but um but for most of my career I

was you know I was a regular Tech guy I

got the PHD in computer science with

with James

uh and you know worked on regular

technology in unrelated to uh but I had

a you know an interlude in my career for

for five years I worked you know

full-time as an artist

uh just doing open source stuff

and I'll show one more uh much lesser

known

example here

of my sort of super talking about

history this is sort of like a super

early attempt at something that is now

like stable diffusion uh so this is a

generative algorithm that takes a an

image as input and sort of uh and

generates stuff that looks uh based on

based you know is modeled after the

inputs and so

as you can see the the here's an attempt

at doing hands it's it still has the

same problems after all these years

oops sorry

there we go

uh yeah this was this was 1993 though so

and so all the these were uh I'll

predate you know uh the adoption of

neural networks we we have today

so

um

but that's

um

so

that's sort of where I'm coming from

somebody who's you know been doing this

uh the open source stuff and the the

generative AI stuff

and uh you know learned about it but

it's all just Prelude to what's been

happening these days and so and you know

and it's Child's Play compared to you

know what the algorithms uh today today

are doing so over the last three years

I've seen a you know an incredible leap

in the computer's ability to understand

you know human language and generate

images

and so what what are the implications of

this

um you know

um you know I would say you know there

we could have like a a private tutor for

every child you know a personal

assistant in everyone's pocket uh you

know new forms of personal expression

you know Photoshop that's easy to use

and just does what you tell it

um

uh interactive fiction you know who

knows uh these are these are just really

like the obvious ideas some of them from

fiction itself but uh you know like

let's say uh George uh melies you know

we're really just at the beginning of

this you know we're still pointing a

camera at a stage and we don't know

what's really going to take off and

where this technology uh is gonna gonna

take us and and what it's unintended

consequences are

um

so

and these you know these these models

these you know language models and image

models contain uh Notions of Truth and

Beauty but you know but who defines uh

truth and uh who defines

Beauty

and you know how can you trust

uh that the the model that you're using

you know serves your interests versus

say the the company that created it

so

um

you know the the models that have been

sort of making headlines are proprietary

products uh from from secretive

organizations and you know there's

issues with trust and accountability

so uh there but there's the good news is

there's an alternative when and that's

open source so and there's four four

parts to that

we need the code for doing for training

the models to be open uh we know we need

the code that runs the models the

inference engine to be open

we need the training data to be open and

uh the model waits to be open

and you know if we have

those four things then you know what

what are the implications

you know if if you can see the data that

goes in the models that allows

independent you know research and you

know third-party audits uh and it can

really help with the trust and Alignment

because you know you can find uh you

know uh what's uh how you can really see

how it works

and the you know liberal licensing

because you know open source makes stuff

free to use and that's going to really

help spread the benefits

of of AI to to everyone

and if the licensing is liberal

which means like allowing commercial use

then

that can motivate investment I mean you

can get like a virtual virtuous a

virtuous uh cycle of of improvement

um

so

and it's really important though that

one of the things that open source

enables is this forking and fine-tuning

of of the models and so you can take a

foundation model and add a little bit

more data and you know change change its

character and change its notion of Truth

and and change it

and that allows you know uh each person

or for every country or every culture or

every identity to create their own

models so instead of having just one

model from one company you can have each

person or each organization or each

group of people you know creating their

own model and you know we're really

lucky the beauty of these pre-trained

models the foundation models is that

making them is extremely expensive but

the fine tuning is actually really easy

and cheap you can just do that in a few

days days

and so you know this

so instead of having you know one notion

of Truth you know we can have you know

freedom of choice and you know bias I

think is really inherent

in the model so we need to just be able

to

um take you know do the one we want and

ultimately every person can have you

know their their own model

so I think that

open source will help a lot with a bunch

of the these sort of hard problems

but there's really one that remains

which is what happens if uh these

powerful tools are used by people for

say nefarious purposes

um

and you know the harm can result and

there's really I think

ultimately

uh I'm optimistic about this and if you

know if you look at the history of

Technology

um I think the ultimately I believe in

in human nature and that the uh there's

more good people than bad people and the

benefits

outweigh the problems and we will

identify the problems and address them

uh you know the best we can as as the

cookie crumbles

so

and uh you know my experience is that

you know open source is part of you know

that solution of making these

uh making this technology work best for

everyone

and so that's uh

the end of my introduction thank you

[Applause]

so good morning

I'm very happy to be here thank you for

the invitation

um uh copyright challenges to generative

AI it's the focus of of my talk today

we could have a really long conversation

about the governance challenges

generally the generative AI is posing

there's a global conversation about this

particular topic obviously

the general public has

embraced uh the chat jpt and many of

these systems that you read about in the

in the paper and a lot of technologists

are enthusiastic about it too but policy

makers are concerned about everything

from uh privacy and cyber security

defamation disinformation just to name a

few so we're just going to talk about

copyright today and a thing to know as

we begin this is that some authors

artists and programmers are very

positive about this uh development and

about copyright as not being an obstacle

and others are very negative there are

three lawsuits right now

challenging generative AI AI on

copyright and related grounds the United

States copyright office is holding some

listening sessions about stakeholders

interests in what how the copyright

office should be thinking about

the main issues that it is addressing

and it will publish a report probably

sometime by the end of the year and the

copyright office actually has a a

website that's all about generative Ai

and so lots and lots of materials there

if you're interested in it I'm going to

only have time to talk about three

questions and I never have to do it

really fast so the question about

whether ingesting copies of copyrighted

works as training data is that an

infringement of copyrights in those

Works obviously pretty much everything

that's out there on the open web is

actually protected by copyright law

unless it was authored by the US

government and that means that there's a

lot of copyrighted stuff out there

um even if people aren't uh exploiting

it in the way that uh that many of the

copyright Industries do there's the The

Works are still protected

um and then uh a second question uh is

whether uh AI generated outputs are um

infringing derivative uh works of

ingested uh content uh two of the

lawsuits against uh stability

um raise this issue as well as the first

one uh and then a third question uh has

to do with removal or alteration of uh

what copyright law calls copyright

management information things about sort

of what is what is the name of the work

uh who's the author and on what terms is

available and the stability cases

raise this issue as does a class action

lawsuit against GitHub and open AI over

the Codex large language model and

co-pilot the the programming assistant

tool that Microsoft is hosting in uh in

the cloud

the briefest thing I can say about

copyright is that um it it attaches

automatically by operation of law from

the first fixation of a work in a

tangible form and the author is the one

who gets those rights and these are the

five major

exclusive rights that copyright grants

to them the right to last for

practically forever and copyright only

protects the original expression of an

author not the ideas not the facts not

the methods there are lots and lots of

unprotected stuff in copyrighted works

and copyrights exclusive rights are

limited by fair use and various other

doctrines and there's a special

copyright like law that makes

intentional removal or alteration of

copyright management information

um illegal if you know that it will

facilitate uh copyright infringement uh

Fair uses are non-infringing and

um so it's a defense to a charge of

infringement and fair use is usually the

defense that's raised when we're talking

about the ingesting of uh copyright

Works uh in uh in as training data and

there are various factors I'm not going

to be able to go into this right now but

I'm certainly happy to talk at the Break

um about these things so is ingesting

um works as training data copyright

infringement or not there are at least

several cases and I'm listing two of

them here uh where the court basically

threw out copyright claims on fair use

grounds so that field put a bunch of his

work up on on the internet uh Google of

course crawls the web and copied it to

index the contents and to let people

find Fields work and Fields that Ah

that's a copyright infringement and the

court said no it's fair use because

Google is using the works not to exploit

the content but to just let people know

that the works uh exist and are out

there and in the author's Guild versus

Google case and Appellate Court rule the

Google's digitization of tens of

millions of in copyright books from

research Library collections was fair

use when it was done for the purpose of

indexing uh the contents and serving up

Snippets in response to search queries

so stability is going to be relying on

these in similar cases to support its

fair use defenses

but there are a lot of people out there

who really don't like the ingestion

without permission and they can't really

easily opt out they weren't paid for uh

the value of their contributions uh and

some part of the concern is that the

images that are being generated through

stable diffusion and others of these

images

generators is that they will compete in

the market with the images that artists

actually are doing and it's in some

sense you're competing against yourself

in some sense at least that's the way

some of the artists think about it and

this cartoon actually is kind of

illustrates how some people think about

generative AI

um

now they're a countervailing

considerations

generally speaking the people who

develop these large language models are

not really interested in the original

expression in a work they're really

interested in essentially understanding

the facts and they kind of think of uh

of of of documents and the like as raw

material for computational uses that are

actually not exploiting the expression

and so that's sort of a way of kind of

looking at it and also generative AI

enables a lot of creative reuses of

things and so what copyright law is

supposed to be doing is promoting the

progress of science or knowledge and

culture and so the people who favor

um

generative AI are thinking that this is

actually a good thing that that

generative AI advances the purposes of

copyright uh outputs as a derivative

Works

um so

authors have exclusive rights to prepare

derivative works but the course have

basically said something can't infringe

the derivative work right if you haven't

just extracted some expression from an

existing work and then put it into a

second work so it's not enough just to

be based upon an existing work you

actually have to have

essentially taken expression and reused

it on the class action lawsuit against

stability and claims that every output

of stable diffusion is an infringing

derivative work and I think that's

actually hard to say under existing

precedence the that that's a sound

result but in general

the text and images that are generated

in response to user props are not going

to be substantially similar in their

expression to the works that were in the

training data and if that's true through

then that's unlikely to infringe the

derivative work right

and

you know there are some examples that

people have shown where you can

essentially like Mickey Mouse and

Superman you do a prompt with them

you're going to get an image that's

going to look like Mickey Mouse or

Superman

of the cases against stability

um Getty is says Hey stability you

ingested 12 million photographs and

captions from our online database that's

actually copyright infringement also

you know there are some images they

claim that are infringing derivative

infringing of the derivative work right

and also there's a copyright management

information claim in this particular

case and here's an example I don't know

whether you can see it but uh

this is getting images and this is the

mangled uh stuff uh and so that's a

basis for claiming uh that there is uh

that that's uh that uh stable diffusion

uh is at least violating the copyright

management information uh law uh and so

with that uh you know these lawsuits are

in very early stages

um it's going to be many years actually

probably until we know what the results

are from the courts but it's actually

good to remember this isn't the first

disruptive technology that's ever been

uh out there um so I like to remember

that piano rolls were a disruptive

technology of the day because the

composers basically said hey that's my

music but it didn't look like a

copyright infringement to the courts and

so it went up to the Supreme Court said

no it's not a copy because it's part of

a machine so copyright law has had to

evolve to kind of recognize that that

you know technology has to be responded

to and sometimes you amend the law in

order to protect copyright owners and

sometimes you know you just let the

technology happen so you know the

recording industry hated

um MP3 players but guess what

um they're like they're legal so anyway

I'll just leave that with the uh for a

discussion and thanks very much for your

time

so I think we have microphones for

questions so if you want to raise your

hand I'll start things off thank you

both for for those really stimulating

comments

um you you referred to a bunch of

lawsuits uh that are ongoing and I and I

was wondering is that how this is going

to play out is it going to be the courts

that figure this out versus kind of a

prospective either regulatory approach

or self-governance by the companies that

are involved in I might be very

interested in both of your thoughts

about those alternatives to just having

the courts make these decisions and

using those precedents well the

copyright office as I said has a clear

intent to put a marker down in terms of

the sort of the major questions that I

was just raising and so they'll have to

they'll have a say about do they have a

good history of moving uh no

um but

um you know they have motivation they

can't do anything by themselves right

they can make recommendations to

Congress to pass legislation

and we all know how functional Congress

is right now so um you know part of what

happens is that when you have a

dysfunctional

Congress and you know the you know the

Biden Administration whatever they might

think about this

um they can't do anything either okay so

lots of things end up in court because

there was a big dispute and there's no

other entity that really can deal with

this I should actually mention that um

one of the things that is important is

that fair use is a limitation on

copyright in the United States there are

a number of other countries that

actually have fair use defenses in their

copyright laws too but most of the other

developed countries have special

exceptions for text and data mining and

so at least for the purpose of ingestion

of copyright works for generative AI the

text and data mining exceptions would

actually kick in and at least for

scientific and non-profit uses

that cannot be overridden by contract

and right now there's a question about

whether people can opt out I remember

the discussion on the first session

about opting out and it's not clear you

can opt out

so Scott on the same issue you had the

four key tenets of Open Source Code open

inference code open data open and I

wonder if you can comment on the data

open I just wanted to hear a little bit

more about how open and how this

interacts with your thoughts about

copyright and

sure the when I say the data open I mean

that the you know the all the the list

of images or texts all the all the

documents that are fed into the training

algorithm is a published list that

anyone can can inspect it's and there's

actually a proposal in the European

Union that um as part of the AI act that

they're considering uh is to have a

requirement that generative AI systems

have to disclose uh what copyright Works

um they ingested so that's kind of

consistent with your kind of preference

for open data in that way yeah so and

then once if if you have a list you can

look and see are are you on it um and

you know copyright lies certainly too

complicated for me to understand I have

no idea uh you know I what what's going

to happen and um uh but I I do believe

that there is you know some some list of

uh you know there's tons of uh public

domain imagery out there there's tons of

information out there there there's all

kinds of Licensing available and I

believe there is a list of data there is

a data set uh that would be uh sort of a

useful and and legal uh and you know we

should we should pursue it you know

together so like let's let's look at the

list and uh

come up with a consensus it's actually

also important to know especially for

any of you out there who are trying to

develop some of these models the

Creative Commons official position

is that using

a Creative Commons works as training

data is actually not a violation of the

Creative Commons licenses now this is in

contrast to the the GitHub lawsuit in

which the four programmers have sued

GitHub open Ai and Microsoft

for breaching open source licenses

because the generative

system that was built on top of a model

that was ingested the open source code

is actually there's claiming it's a

breach of license because there's no

attribution right and so that's actually

an issue that where they at least taking

that position now again

who knows what the courts will do with

that I did want to take some questions

from the audience do we have

I saw this hand first

thank you very much for your

thought-provoking comments uh one of the

points you made were about uh truth and

Beauty right

crude

coming from a scientific background you

kind of think about it as objective

evidence something which is backed by

objective evidence but beauty is more

subjective

it can be interpreted by the Iowa to

Beholder right so can you elaborate on

that in in terms of how can we make sure

that

we don't get into the world of

alternative facts

but maintain one notion of truth if it

makes sense

uh you know there's certainly scientific

uh truth is maybe uh the the easier part

like at least we have like you know the

scientific method that we have a way

that we can agree a process to figure

out what that is obviously it's executed

by people and imperfect

um but there's you know there's there's

all kinds of truths uh you know there's

historical truths uh and there's you

know personal truths and I just really

don't see us uh coming up with a single

definition for that and so that's why we

need you know the ability for people to

fine-tune their own versions of the

models and

ultimately the you know

Define their Define their own truths and

yeah those could be if you disagree with

them you you might uh call it

misinformation or alternative facts I

think uh

you know uh but I you know I I just

don't see uh any way of uh you know uh

uh regulating a single answer there and

I think it's better to uh you know allow

um you know diversity of those terms A

diversity of Truth and Beauty

um it's so that's that's uh that's

that's uh how I that's my position

we have one two and then if we're really

lucky three

um thank you so much for uh your

comments I have a question that sort of

um touches on both aspects of uh

generative AI from the artist's

perspective as well as when you look at

copyright infringement and so I don't

know if everyone is familiar with uh the

new trend of Music covers by generated

AI machines

um but more so coming from the consumer

um listening to a piece of music and

wondering what it would sound like if a

different artist sang it

um and I know Universal Music put out a

statement saying they would pursue legal

action against those types of music

covers

um but when another human makes a cover

of a musical piece it's not illegal

right there they have the right to

creatively Express themselves

um on a piece of music and so I'm I'm

wondering from the artist's perspective

as well as from a legal perspective how

can we always even possible to regulate

this intersection of generative AI meets

consumers that are paying to have access

to certain types of music as well as

copyright and intellectual property from

the artist's perspective

degree because we do have a panel in the

afternoon that addresses that too so

thank you so much for the great question

well I think that to the extent that

some software tool

enables people to you know do their

karaoke as a as a kind of

generate the music that you want to hear

that's going to be something that

Universal probably won't know about and

won't care about

it's kind of like fan fiction kind of

stuff which mostly has been either

considered fair use or been tolerated by

people but you try to go you try to go

public with that you try to

commercialize it and Universal I

guarantee Universal will be there and

one thing that actually is going to be I

think more significant

in the generative AI space then it has

been in other contexts and that is right

of publicity so there is this law it's

really state law which basically says

that people have a right to control the

use of their names and likenesses and

some for kind of commercial purposes so

if somebody takes a picture of you and

then uses that as an advertisement for

some other product that's actually a

violation of right of publicity and that

the the Drake song that went viral is

something where I think there was a

right of publicity issue in that case I

mean I didn't turn into litigation

because it just got taken down really

fast because I can tell you that

Universal makes its voice heard really

loud with anybody that's hosting stuff

like that so it got taken down pretty

fast but it went viral really really

fast too and it's certainly not going to

be the only time when a sample of

someone's voice is used to

then make a music that sounds like the

that's them when it's not so they're

also laws against false representation

and so there are kind of some existing

laws other than copyright that um that

may come may have some bearing on that

we are now entering the lightning round

all right really quickly I I really

appreciate um your your response got to

the question about scientific truth and

I think it's really important to push

back on this privileging of scientific

truth because it's you know I think from

this audience it is just as ambiguous as

the definitions of beauty I think I

wanted to ask you quickly if you have

you've been really unique in giving away

a huge amount of your your work and even

without asking for credit or attribution

so I'm curious if you have any regrets

or advice on that and uh and really

quick to Pam if we have time but uh is

it has this recent Warhol versus

Goldsmith's decision changed anything in

in regard to AI

um I I have uh I don't have any regrets

I mean I've lived a Charmed existence

I've been incredibly lucky

um and it's it's hard to untangle you

know like what what would have happened

um but so I think it's it's it's I think

it's working great I would recommend it

and um

yeah I I don't know uh

let's go let's go to the next one thank

you just

I'm sorry

yeah go so short version

um the United States Supreme Court just

decided that that a commercial license

of an image that

Andy Warhol created in 1984

so the commercial license in 2016

was

not uh a fair use of that image they

didn't reach the question about sort of

how far and how whether the works when

they were created were infringing or not

they just basically said that's the that

Goldsmith had abandoned those claims and

so really really narrow issue as a

commercial license

of that particular image Court didn't

say it was uh it was necessarily

infringing but said that it wasn't the

transformative enough use of the image

so you know there's some dicta in that

opinion that a lot of people who hate

fair use are going to just love but for

them for the most part really narrow

narrow ruling

we do have some questions from the from

the audience remotely and but many of

the issues have at least been touched

remote uh and glancing blows

um I wanted to just ask you Pam just uh

to speculate or sort of what your

assessment is of the um viability of

fair use claim given that the fourth

factor is uh maybe increasingly like uh

in question uh

since the AI companies now have created

this booming market for data and the

value of that data now is sort of widely

recognized and a lot of companies who

have access to a lot of data are

creating licensing agreements for the

use of that data in training AI does

that change your you know guess I guess

about how courts will fall on the fair

use question well I think

I think that uh Getty Images lawsuit

against stability

um that's a that's a that's a more kind

of

um important consideration because

um getting images says I don't I I'm

completely happy for you to use uh my

images and the captions uh as uh

training data but I have a licensing

program for that and so it's not fair

use the class action lawsuit I think is

quite a different thing because the the

three artists that claim to be the class

plaintiffs

are trying to represent all of the

artists of all the visual images in the

world that

um that were ingested you can't get a

license there so it seems to me that

that the fair use claim is going to be

stronger where it's infeasible to

um to do it I don't know that the

existence of look every time somebody

brings a lawsuit they're basically

saying I

lice I have licenses okay or I'm willing

to license this particular thing so the

fact that you want to license the fact

that you have a licensing program

doesn't actually necessarily mean that

it's not fair use I am not going to

predict here but certainly the Getty

Images case has a a stronger response to

the fair use claim than the the Anderson

class action case

well I want to thank the panelists and

thank the uh for the questions

[Applause]