Name: Navigating the legal and ethical implications of AI
Uploaded: 2024-02-19T11:57:31.799Z
Duration: 325.86
Description: The use of AI is still in fairly murky waters when it comes to legal and ethical implications, leaving marketers wondering how to use this technology safely. Christopher Penn, Chief Data Scientist at Trust Insights, discusses the vital considerations marketers, from copyright to disclosure.

Ethical considerations v3-.mp4

Ethical considerations v3-.vtt

English

{{00:00:07}}
For the ethical considerations
that marketers should be looking

{{00:00:09}}
at, there's two branches.

{{00:00:10}}
There's ethics and then there's
law, and they're separate and

{{00:00:13}}
they're both equally important.

{{00:00:15}}
Marketers need to understand
the law around AI.

{{00:00:17}}
So the first part is, for example,
what models are you using?

{{00:00:20}}
And were those models constructed with
works that may or may not have,  used

{{00:00:25}}
other people's works without permission?

{{00:00:27}}
Depending on where you are in
the world, the law is different.

{{00:00:29}}
So if you are in Europe, for example, the
EU has ruled that models that leverage

{{00:00:34}}
other people's works for training
data have violated their copyright.

{{00:00:36}}
And so there are a bunch of AI services
that are not available in the EU.

{{00:00:40}}
In Japan, Japan's laws say that training
a model does not violate copyright, and

{{00:00:46}}
so,  again, there's, they're trying to
spur tech development there, and as a

{{00:00:50}}
result, they're making the laws very
permissive, and in the United States,

{{00:00:53}}
that has not been settled into law yet.

{{00:00:55}}
On the flip side, generated works,  for
example, in the United States, generated

{{00:01:00}}
works by machines cannot be copyrighted.

{{00:01:01}}
In 2018, there was a famous
case, Nerudo versus Slater.

{{00:01:04}}
When a chimpanzee took a selfie and
the photographer tried to copyright

{{00:01:08}}
it and the courts ruled that the
photographer did not do the work

{{00:01:11}}
of taking the photo, the chimpanzee
did and chimpanzees are not human.

{{00:01:15}}
They cannot hold copyright.

{{00:01:16}}
And so that public that image
is automatically in the public

{{00:01:18}}
domain when chat GPT or stable
diffusion or Dolly creates a work

{{00:01:23}}
if it is a machine made work.

{{00:01:25}}
It cannot be copyrighted.

{{00:01:26}}
So if you are a marketer and you've put
up a blog post on your blog that was

{{00:01:29}}
generated by a machine, a competitor
can copy paste the whole thing.

{{00:01:32}}
And there are, there are
no legal repercussions.

{{00:01:34}}
You cannot enforce your copyright on it.

{{00:01:36}}
No matter what a big part of what
needs to be settled is whether or

{{00:01:40}}
not the law will change so that.

{{00:01:43}}
Machine generated works can be copyrighted
and right now that is still unsettled.

{{00:01:48}}
I have a feeling that big content
companies like, old Disney,  will

{{00:01:54}}
successfully lobby eventually to have
machine generated works because they're

{{00:01:57}}
not going to ignore billions of dollars.

{{00:01:59}}
On the ethics side, the question of
disclosure and transparency is important.

{{00:02:04}}
Again, going back to the legal side, if
you disclose the use of AI for certain

{{00:02:08}}
parts of say your marketing collateral.

{{00:02:11}}
Then by extension, you are also
reinforcing your copyright claim

{{00:02:15}}
over the human generated parts.

{{00:02:16}}
So disclosure is important, not just to
inform consumers about what is and is not

{{00:02:20}}
machine generated, but also to say, hey,
in this whole part here, that's ours.

{{00:02:24}}
You, you can't use it.

{{00:02:25}}
Like under the law, you, you could
theoretically reuse this blog post image

{{00:02:29}}
header because it was machine made,
but our content, you can't reuse that.

{{00:02:32}}
So you're, you're.

{{00:02:33}}
Making the boundaries very
clear so you can enforce your

{{00:02:36}}
own copyrights in general.

{{00:02:39}}
Disclosure is better than not in
general, telling people, Hey, this

{{00:02:42}}
was made in part or in full by a
machine is better not because at the

{{00:02:46}}
end of the day, the consumer does care
about transparency to some degree.

{{00:02:51}}
Certainly we have seen plenty of use cases
where Security and and safety and stuff.

{{00:02:58}}
Disclosures exist.

{{00:02:59}}
People don't care.

{{00:03:00}}
They will hand over their personal
data for a cookie, literally.

{{00:03:03}}
And so disclosure does not
harm us in our in our use of A.

{{00:03:08}}
I.

{{00:03:08}}
Because again, the consumer doesn't care.

{{00:03:09}}
The consumer just wants.

{{00:03:11}}
Whatever it is thereafter, whether
it's machine made or not, even

{{00:03:14}}
B to B marketing, the buyer
doesn't care if this white paper

{{00:03:17}}
was written by a machine or not.

{{00:03:18}}
The buyer cares.

{{00:03:19}}
Did this answer my questions
that I know what to buy?

{{00:03:22}}
What's happening right now?

{{00:03:23}}
We are on a very, very fast track towards
all of us having a personal assistance.

{{00:03:28}}
Microsoft copilot, for example,
will be your personal assistant

{{00:03:30}}
within the Microsoft Office system.

{{00:03:32}}
Google duet will be the same in
the Gmail and Google ecosystem.

{{00:03:35}}
We're going to have these AI
driven Microsoft Agents as personal

{{00:03:39}}
assistants throughout our life.

{{00:03:41}}
And so the average consumer will
be like, okay, I'm happy having

{{00:03:44}}
this machine do this for me.

{{00:03:45}}
I know it's a machine and
I'm comfortable with it.

{{00:03:48}}
If it saves me time, saves me
money or helps me make money.

{{00:03:51}}
And so disclosing it now as part of
your ethics is a good idea because It

{{00:03:56}}
will turn a point where people will
expect the level of quality coming

{{00:04:01}}
out of machines and will want that.

{{00:04:03}}
We'll say like, I would
rather deal with a machine.

{{00:04:06}}
If you have a choice between waiting
in line for 90 minutes at your local

{{00:04:10}}
motor vehicles registry or dealing with
a machine to get your thing done in 90

{{00:04:14}}
seconds, you're going to pick the machine.

{{00:04:16}}
The question that people have to ask
in terms of ethics is at what point

{{00:04:20}}
in the process from beginning to end
of any marketing task or any business

{{00:04:24}}
task period should you be using AI or
should you not be and if you should

{{00:04:29}}
or should not be under what are the
conditions that you're going to disclose

{{00:04:32}}
it and what are the conditions which
you would be prohibited from using it.

{{00:04:37}}
And those are things that companies need
to figure out sooner rather than later.

{{00:04:40}}
Each company individually has to
figure out where their positions are.

{{00:04:44}}
It's going to vary.

{{00:04:45}}
For example, in higher education,
higher education is in a existential

{{00:04:49}}
crisis at the moment because artificial
intelligence can write a better

{{00:04:54}}
term paper than any college student.

{{00:04:56}}
Period.

{{00:04:56}}
End of story.

{{00:04:57}}
And the tools that can detect machines
are about 50 percent accurate.

{{00:05:02}}
It's a coin flip.

{{00:05:03}}
As a result, professors are
like, well, what do we do?

{{00:05:06}}
Some progressive professors are
like, we're going to let students

{{00:05:09}}
use this and then have them sort of
critique what the machines put out.

{{00:05:13}}
And regressive companies and
institutions are saying, AI.

{{00:05:16}}
Which, again, good luck
detecting it, guys.

For the ethical considerations
that marketers should be looking

There's ethics and then there's
law, and they're separate and

Marketers need to understand
the law around AI.

So the first part is, for example,
what models are you using?

And were those models constructed with
works that may or may not have,  used

Depending on where you are in
the world, the law is different.

So if you are in Europe, for example, the
EU has ruled that models that leverage

other people's works for training
data have violated their copyright.

And so there are a bunch of AI services
that are not available in the EU.

In Japan, Japan's laws say that training
a model does not violate copyright, and

so,  again, there's, they're trying to
spur tech development there, and as a

result, they're making the laws very
permissive, and in the United States,

On the flip side, generated works,  for
example, in the United States, generated

In 2018, there was a famous
case, Nerudo versus Slater.

When a chimpanzee took a selfie and
the photographer tried to copyright

it and the courts ruled that the
photographer did not do the work

of taking the photo, the chimpanzee
did and chimpanzees are not human.

And so that public that image
is automatically in the public

domain when chat GPT or stable
diffusion or Dolly creates a work

So if you are a marketer and you've put
up a blog post on your blog that was

generated by a machine, a competitor
can copy paste the whole thing.

And there are, there are
no legal repercussions.

No matter what a big part of what
needs to be settled is whether or

Machine generated works can be copyrighted
and right now that is still unsettled.

I have a feeling that big content
companies like, old Disney,  will

successfully lobby eventually to have
machine generated works because they're

On the ethics side, the question of
disclosure and transparency is important.

Again, going back to the legal side, if
you disclose the use of AI for certain

Then by extension, you are also
reinforcing your copyright claim

So disclosure is important, not just to
inform consumers about what is and is not

machine generated, but also to say, hey,
in this whole part here, that's ours.

Like under the law, you, you could
theoretically reuse this blog post image

header because it was machine made,
but our content, you can't reuse that.

Making the boundaries very
clear so you can enforce your

Disclosure is better than not in
general, telling people, Hey, this

was made in part or in full by a
machine is better not because at the

end of the day, the consumer does care
about transparency to some degree.

Certainly we have seen plenty of use cases
where Security and and safety and stuff.

They will hand over their personal
data for a cookie, literally.

And so disclosure does not
harm us in our in our use of A.

Because again, the consumer doesn't care.

Whatever it is thereafter, whether
it's machine made or not, even

B to B marketing, the buyer
doesn't care if this white paper

Did this answer my questions
that I know what to buy?

We are on a very, very fast track towards
all of us having a personal assistance.

Microsoft copilot, for example,
will be your personal assistant

Google duet will be the same in
the Gmail and Google ecosystem.

We're going to have these AI
driven Microsoft Agents as personal

And so the average consumer will
be like, okay, I'm happy having

I know it's a machine and
I'm comfortable with it.

If it saves me time, saves me
money or helps me make money.

And so disclosing it now as part of
your ethics is a good idea because It

will turn a point where people will
expect the level of quality coming

We'll say like, I would
rather deal with a machine.

If you have a choice between waiting
in line for 90 minutes at your local

motor vehicles registry or dealing with
a machine to get your thing done in 90

seconds, you're going to pick the machine.

The question that people have to ask
in terms of ethics is at what point

in the process from beginning to end
of any marketing task or any business

task period should you be using AI or
should you not be and if you should

or should not be under what are the
conditions that you're going to disclose

it and what are the conditions which
you would be prohibited from using it.

And those are things that companies need
to figure out sooner rather than later.

Each company individually has to
figure out where their positions are.

For example, in higher education,
higher education is in a existential

crisis at the moment because artificial
intelligence can write a better

And the tools that can detect machines
are about 50 percent accurate.

As a result, professors are
like, well, what do we do?

Expert Insight

The use of AI is still in fairly murky waters when it comes to legal and ethical implications, leaving marketers wondering how to use this technology safely. 
Christopher Penn, Chief Data Scientist at Trust Insights, discusses the vital considerations marketers, from copyright to disclosure. 

Content Marketing

Ethics

Marketing insights

Marketing strategy

thumbnail_Christopher-Penn_CMWorld-2022-copy_491429_1662484410.png

xsmall_Christopher-Penn_CMWorld-2022-copy_491429_1662484410.png

Christopher-Penn_CMWorld-2022-copy_491429_1662484410.png

Christopher Penn

Chief Data Scientist

Penn provides insights into the legal implications of marketers using AI, with examples from across the globe.

Knowing your laws

Penn looks at the importance of disclosing the use of AI in any marketing materials.

Disclosure & Transparency

Learn the criteria companies should be checking against themselves when determining where and when to use AI. 