The world’s leading publication for data science, AI, and ML professionals.

A quick reflection on some ethical implications of creative AI

AI is increasingly being applied to more creative areas, raising concerns about the protection of intellectual property.

Photo by Nick Morrison on Unsplash
Photo by Nick Morrison on Unsplash

Disclaimer: I am not a lawyer, and therefore, this article should not be used as legal advice, so take it as a personal opinion of an experienced observer of emerging technologies and market shifts.

Recently I started to analyze a trend taking AI towards a more "creative" space with the integration with human creativity.

As content generation capabilities of AI increase, it starts to be applied more often in the more diverse areas, raising concerns about how its content can be protected as intellectual property.

Let’s consider for a moment all the industries and sectors that rely on intellectual property to generate and capture value. This set includes software, electronics, consumer products, pharmaceuticals, medical devices, video games, film and television, journalism and news media, and music.

The market value of these industries and the individual players’ margins are closely tied to IP protection.

As AI contributes more and human beings contribute less to the value generated, we can expect to see more significant ambiguity around whether IP is protected soon.

In the near term, it doesn’t look like the policy environment is changing much. The current trend appears to be reaffirming human beings’ role as inventors and authors and patent and copyright offices in protecting endeavors of human creativity against AI-generated content.

Photo by Sam Moqadam on Unsplash
Photo by Sam Moqadam on Unsplash

AI-assisted versus AI-generated content

In AI-assisted endeavors where human beings are making a creative contribution – generally the current state of the art – the AI would be considered a tool that doesn’t need to be listed as a co-inventor, and those team members would be listed as inventors and authors.

But an important question is not whether AI can be a producer but rather whether AI outputs can be copyrightable.

An AI that augments specific skills usually would not be significantly different. Still, when we have more AI capabilities to be inventors or authors with a relevant academic or corporative role, patents will rise. Hopefully, a distinction between AI-assisted and AI-generated output with taking form.

If we start to imagine the scenario where specifically AI-generated works – where the AI finds the data, trains itself, and "creates" a specific output without any human intervention – we can imagine how it might put a strain on existing IP rules.

In the case of AI-generated inventions, we’ll see human beings as the middleman making dubious claims about inventorship and operate under a state of more significant uncertainty regarding the validity of the IP protection.

We’ll encounter some odd situations where AI’s output will dramatically surpass what human beings can create – not just in pieces like "sweat of the brow" but in real Shakespeare-level novelties – but not merit IP protection under the strictest interpretation of the law.

Raising an ethical dilemma here… who owns the rights and responsibilities, for example, if we have an AI that is still in operation and autonomously generating inventions but all the original team members who developed the AI are dead?

In this case, the editor or the team’s company would still own the AI. Still, any claims of inventorship by its employees in a patent application would be a stretch, no?

There’s also the question of who holds the liability for an autonomous AI engaging in questionable acts such as patent infringement if there’s no prominent person or company in control of such an AI.

Answering about who would be considered responsible for the information written by an AI system, Idoia Ana Salazar García, **** Professor at the School of Communication of San Pablo CEU University and Principal Investigator for Social Impact of Artificial Intelligence and Robotics (SIMPAIR), affirms that the manager would be responsible for media matters. Eventually, the machine will be like another employee, giving the journalist suggestions on which angle to cover.

That is why we must educate media managers and editors on how algorithms’ decisions can have a significant social impact.

Of course, copyright is a different case than patents. Currently, the US only registers copyrights for human-generated works (even considering AI assistance). Simultaneously, other countries such as the UK, India, New Zealand, South Africa, and Hong Kong have established a framework and precedent for protecting computer-generated works, where the author still needs to be human. Yet, authorship is assigned to the person who arranged for the creation of the work.

Photo by Possessed Photography on Unsplash
Photo by Possessed Photography on Unsplash

But who will protect AI’s rights?

How AI itself is protected is a separate question from how its output is protected.

There are aspects of AI that can be protected through patents and copyrights. There are some elements – such as datasets and algorithms – that are well-suited to being trade secrets.

As AI becomes increasingly valuable, we will see a significant pick-up in attempts to steal sensitive AI technologies. AI companies, in turn, will need to defend their crown jewels with cybersecurity solutions and programs.

There’s also the question of whether an AI can be trained on protected content like music, which is still not fully addressed in the law.

While copyright owners generally hold the rights to derivative works, the use of protected data to train algorithms may fall under fair use if the trained algorithms don’t reduce the original work’s value and the trained AI’s material is minimal.

By contrast, if the AI retains a database of protected material for purposes of comparison (e.g., facial recognition for surveillance), that might not be ethical to be used, as affirmed by the US Intellectual Property Owners Association (IPO) in 2020.

Some countries are exploring explicit exemptions for using copyrighted material in training AI – Japan has already updated its laws.

Photo by Rod Long on Unsplash
Photo by Rod Long on Unsplash

Conclusion

The possibilities in this field are endless, and we are only at the beginning of this creative revolution.

At this moment, there aren’t any upcoming policies to inhibit or influence AI authorship or patent offices as long as they stay within the guidelines set by current regulations.

It means that, in the case of fully AI-generated inventions, we’ll operate under a state of more significant uncertainty as to the validity of the IP protection for a long time still, living with the open question:

Who owns the rights and responsibility of AI-generated inventions?

One more thing…

If you want to read more about Creative Ai and how you can start learning about it, you probably want to read the following articles too:

An overview of Creative AI

A Simple Approach to Define Human and Artificial Intelligence

The best free courses to learn AI, ML, and Data Science today.

Also, I’ve just published a new ebook on Amazon, and I’m already working on publishing some others this year… let’s keep in touch, and let’s do it together.

Jair Ribeiro

References

  1. The use of AI in data journalism: what are the ethical …. https://sciencemediahub.eu/2019/10/08/the-use-of-ai-in-data-journalism-what-are-the-ethical-implications/
  2. Artificial intelligence Ethics guidelines for developers …. https://www.emerald.com/insight/content/doi/10.1108/JICES-12-2019-0138/full/html
  3. Exploring End User Licensing Agreements – Alex John Lucas …. https://alexjohnlucas.com/type/eula
  4. An Overview of Creative AI – https://medium.com/towards-artificial-intelligence/an-overview-of-creative-ai-3fb2e6d2706b
  5. AI, Ethics and Copyright – Hugh Stephens Blog. https://hughstephensblog.net/2020/03/16/ai-ethics-and-copyright/
  6. Artificial Intelligence | Shaping Europe’s digital future. https://ec.europa.eu/digital-single-market/en/artificial-intelligence
  7. Research for CULT Committee – The Use of Artificial …. https://www.europarl.europa.eu/RegData/etudes/BRIE/2020/629220/IPOL_BRI(2020)629220_EN.pdf

Related Articles