Artificial Intelligence: Power Imbalanced

by Jacob Metzger, Associate Member, University of Cincinnati Law Review Vol. 91

I. Introduction

In the last year, use and commercialization of artificial intelligence has exploded.¹ Artificial intelligence (“AI”) is the use of machine learning technology to perform tasks, make rules, or predict based on existing datasets.² The recent boom of AI in art, law, and education has raised a multitude of problems which leave modern law and society unprepared to protect personal safety and property interests.³

This article argues that, while AI may be disruptive, most of AI is beneficial, aside from circumstances that endanger a person’s right to privacy and security. Part II of this article highlights the disruption and power of AI in art, the legal field, personal security, and in the right to personal autonomy. Additionally, Part II focuses on the potential benefits of AI to explaining complex legal topics. Part III details the duality of regulating AI––that it’s necessary for legislatures to penalize harmful uses of AI while also leaving the innocuous usage unregulated. Part IV includes AI generated blog posts to demonstrate the effectiveness of AI in explaining complex legal topics.

II. Background

A. AI Generated Art

In the creative sphere, artists have struggled to compete with AI due to AI’s low cost and advanced capability to analyze and replicate an artist’s style.⁴ AI programs can create art of anything, in any style, at a low cost, almost instantaneously.⁵ Ethical and legal problems arise from this practice.⁶ First, AI generated images are of a similar quality as the art made by artists.⁷ AI generated works have won art competitions when pitted against art made by artists.⁸ Second, for AI to function, it must have artwork to base its algorithm on, and therefore, it must analyze and dissect the style of various artists.⁹ A variety creators have legally challenged AI’s capability to do this under copyright law, but have yet to prevail.¹⁰ Because under copyright law, the use of images to train the AI is likely to be deemed as fair usage, these suits are likely fail unless the law changes.¹¹ Third, AI generators pump the market full of artwork that is not copyrightable, thereby oversaturating the field.¹² Thus, internet users can, without consequence, take and use AI generated images without fear of civil liability.¹³

Copyright law escribes its protections to “original works of authorship fixed in any tangible medium of expression, . . . from which they can be perceived, reproduced, or otherwise communicated . . . .”¹⁴ Historically, the Supreme Court has further refined the term “authorship” to encompass human authors and explicitly refrained from encompassing images generated solely by machines.¹⁵ Recently, the United States Copyright Office reiterated this definition in its cancellation of a registration for images of a comic book––removing the protection of copyright law from the AI generated images used in the comic.¹⁶ Thus, AI generated works will dominate and change the art industry.¹⁷

B. AI in the Legal Field

Like all fields, the legal field will be subject to further development in AI which will change the profession; however, lawyers have various gatekeeping mechanisms to protect legal practice from interference by AI.¹⁸ In the emergence of eDiscovery, AI is commonly used to sift through massive amounts of data that would be too voluminous for a regular person to analyze.¹⁹ Additionally, other websites offer AI for transactional legal work that generates work product like contracts or operating agreements.²⁰ This use of AI is commonplace and widely accepted by the legal community.²¹ However, other uses of AI have been met with resistance due to laws regulating who, or what, can provide legal advice.²² In California, the creator of the application “DoNotPay” announced that the company’s AI chatbot would represent a defendant in court.²³ The creator planned on having a defendant in traffic court wear an earpiece to listen to court proceedings then an AI would generate legal arguments for the defendant to recite, live and in court.²⁴ Various state bar officials threatened to prosecute the creator with up to six months in jail because the AI would be practicing law and not an attorney barred in the jurisdiction.²⁵ Subsequently, due to the threats of imprisonment, he refrained from using the AI.²⁶

C. AI in Deepfakes

Recently popularized deepfakes generated through AI have increased concerns regarding security and the right to privacy.²⁷ Deepfakes use deep learning artificial intelligence to replace the likeness of one person with another in video and other digital media.²⁸ Some uses of deepfakes are relatively harmless, such as comedic spoof videos of politicians and businessmen playing video games together; however, these videos can be used in dangerous ways that the law has not yet adapted to.²⁹ First, deepfakes could further amplify the spread of fake news.³⁰ Deepfakes can present a video image of a person that is near indistinguishable from the person it is portraying.³¹ To the unknowing viewer, this can be used to enrage, radicalize, and stoke fear in others, so the deepfake creator can manipulate them to action.³²

Deepfakes can also be used to violate a non-consenting person’s right to privacy.³³ For instance, celebrities and influencers have become victim to non-consensual deepfake pornography that portrays their likeness.³⁴ Over 90% of all deepfake porn is nonconsensual and the victims are overwhelmingly women.³⁵ Currently, the law has very few avenues to prevent and punish bad faith actors that create, program, and promote this deepfake porn taking the features of a non-consenting person.³⁶ Some state revenge-porn statutes may or may not cover AI-generated deepfake porn; the conclusion is mixed as to whether courts would expand the statute to generated images.³⁷

D. AI in Explanation

AI creates a multitude of benefits with its ability to explain and break down complex topics. For instance, AI can scan the internet and break down complex topics into simple paragraphs.³⁸ In the education system, this presents new problems for teachers and educators with regard to plagiarism; on the other hand, it presents an interesting and efficient opportunity to explain a topic.³⁹ Using various AI, an individual can break down complex learning subjects that can be effective in educating a person about a number of topics.⁴⁰

In the legal field, this could chip away at the barriers to accessing justice or legal information.⁴¹ Various blog writers are using ChatGPT––a recently developed, hyper-intelligent chat bot––to explain complicated legal concepts.⁴² Further, ChatGPT offers the option for users to request the information in more comprehensive, simpler, or artistic styles.⁴³ A user could request that a prompt be explained in the language of a three year old, a cowboy, a hippie, or in formal article style––there are near-limitless opportunities for accessibility to comprehendible legal knowledge, even for hippies and three-year-olds.⁴⁴

III. Discussion

AI has revolutionized and changed the world, and while some uses of AI are beneficial to society as a whole, the legislature and regulatory bodies need to act appropriately and specifically to curtail harm that will drastically increase as AI is left unmonitored.⁴⁵

A. Death of the Artist, Life for the Attorney

Unfortunately, even with sufficient legal action and regulation, it is likely that AI-generated works will rise to prominence over the works of artists.⁴⁶ The artist’s prevalence in society will dwindle to that of a human-cashier in a world of self-checkouts, given that consumers value efficiency and cost-effectiveness over experience.⁴⁷ If the legislature vested the protection of copyright law in AI-generated works, the price of AI generated work would still be lower than that of an artist.⁴⁸ By giving protection to AI-generated works, the law would only prevent unauthorized copying of AI-generated works; thus, vendors of images would still be able to produce and sell their product.⁴⁹ The solution does nothing to prevent AI from simply being a cheaper option for a similar product.⁵⁰

The legal field, on the other hand, will likely be safe from AI control over jobs, due to the gatekeeping mechanisms in place.⁵¹ In stark contrast to starving artists, attorneys legislate, interpret, and enforce the laws dictating who can practice law and they have the power to protect their profession.⁵² While tools can exist to make their lives easier, attorneys have a legal capability that no AI can legally possess: the ability to practice the law and, more importantly, the influence to pass laws and prevent others from practicing law.⁵³ Most fields, like those of the artist, the grocery-store clerk, or the truck driver, do not have access to the power and influence that lawyers possess to protect their trade and will suffer as a result.⁵⁴

B. Deepfakes

Significant legislation is needed to criminalize deepfake media, specifically at the federal level.⁵⁵ Currently, no federal law criminalizing the action exists, and only New York, Georgia, Virginia, and California have laws that specifically address deepfakes in their revenge porn statutes.⁵⁶ California and New York law institute a private right to action, permitting the claimant to sue for damages based on a sexually explicit depiction of an individual.⁵⁷ Georgia and Virginia law institute criminal penalties, which make the creation of deepfake porn a misdemeanor.⁵⁸ Due to the seriousness of the action and the harm considered, both penalties should be instituted federally.⁵⁹ As such, to protect the victim and deter the conduct, the federal government should institute a private right of action and a government penalty to prevent porn deepfakes from destroying the lives of others.⁶⁰ Without clear action to punish and deter those who take the likeness of another for deepfake porn, the power imbalance created by AI, mainly against women, will continue to perpetuate.⁶¹

C. AI for Explanation

For applications like ChatGPT, no update to the current laws is necessary. If the legislature were to expand copyright protection to non-humans, then they would be creating a barrier to legal access.⁶² Expanding copyright protection to applications like ChatGPT would prohibit unauthorized users from publicizing ChatGPT responses to the public; therefore, creating an obstacle to simple, user-friendly, and understandable knowledge.⁶³ To illustrate, Part IV of this article contains two examples of ChatGPT generated responses to legal questions in the style of a law review article and in the style of a third grader. As demonstrated by the example, ChatGPT is able to communicate complicated legal topics even to children; thus, has an immense amount of value in education and power in jumping the hurdles of legal access.

IV. Examples of Chat GPT Capabilities

A. Chat GPT Prompt #1: “Write a law review article about the perils of AI-Generated images on artists and the copyright ramifications of AI-Generated images”⁶⁴

Artificial Intelligence (AI) has made significant advancements in recent years, including the ability to generate images that are indistinguishable from those created by human artists. While this technology has many practical applications, it also poses a threat to the livelihoods of artists and raises complex questions about copyright ownership. This article explores the perils of AI-generated images on artists and the copyright ramifications of AI-generated images.

Perils of AI-Generated Images on Artists

AI-generated images can pose a significant threat to the livelihoods of artists. One of the main benefits of AI-generated images is that they can be produced quickly and at a low cost. This means that companies and individuals who previously relied on human artists to create images for them may now turn to AI-generated images instead. This could result in fewer job opportunities for artists and reduced demand for their work.

Furthermore, AI-generated images could lead to a devaluation of original artwork. If AI-generated images become ubiquitous, consumers may be less willing to pay a premium for original artwork created by human artists. This could make it even more challenging for artists to make a living from their work.

Another challenge for artists is the potential for AI-generated images to be used without permission or attribution. As AI-generated images can be produced so quickly and cheaply, it is easy for individuals or companies to use them without obtaining the necessary permissions or giving credit to the original artist. This could result in the original artist losing control of their work and not being appropriately compensated for its use.

The copyright ramifications of AI-generated images are complex and raise many questions about ownership and authorship. One of the main challenges is determining who owns the copyright in an AI-generated image. Is it the individual who programmed the AI to generate the image, the company that owns the AI, or the original artist who provided the source material for the AI to learn from? This question is particularly relevant when an AI-generated image is sold or licensed for commercial purposes.

Another question is whether an AI-generated image can be protected by copyright at all. In many jurisdictions, copyright law requires that a work be original and the product of human creativity to be eligible for protection. However, AI-generated images are not created by human artists, so it is unclear whether they meet this requirement.

There is also the question of whether an AI-generated image can infringe on the copyright of an existing work. For example, if an AI is trained on a dataset of copyrighted images and produces a new image that is similar to one of those copyrighted images, does it constitute copyright infringement? The answer to this question will depend on various factors, including the level of similarity between the two images and whether the new image can be considered a transformative work.

Conclusion

AI-generated images have the potential to revolutionize the art world and many other industries. However, they also pose significant challenges for artists and raise complex questions about copyright ownership and authorship. As this technology continues to evolve, it is crucial that policymakers and the legal community address these challenges to ensure that artists are appropriately compensated for their work and that copyright law remains relevant in the age of AI.⁶⁵

B. ChatGPT Prompt #2: “In the language of a 3 year old explain the perils of AI-Generated images on artists and the copyright ramifications of AI-Generated images.”⁶⁶

Okay, so you know how sometimes we use computers to draw pictures? Well, some very smart people have made computers that can draw pictures all by themselves! But, even though the pictures look really cool, they can be a problem for artists who make pictures for a living.

See, when an artist makes a picture, they own that picture and no one else can use it without their permission. But with AI-generated images, it’s not always clear who really made the picture. Was it the computer that made the picture, or was it the person who programmed the computer to make the picture?

This can be a big problem because if someone else starts using the AI-generated image without the artist’s permission, it’s like they’re stealing something that doesn’t belong to them. And it’s hard to tell who should really be in charge of deciding if the image can be used or not.

So, it’s important for artists to be careful and make sure that they’re protecting their work, even if it was made by a computer. And for people who want to use AI-generated images, they need to be respectful and get permission from the artist before using them.⁶⁷

V. Conclusion

The law must consider and adapt to the newest technological hurdles presented by AI. Current areas of the market, such as artistry, will not be able to overtake the sway of AI.⁶⁸ In addition, extensive regulation or legislation would likely hinder the various benefits of AI.⁶⁹ Therefore, lawmakers must tailor regulations, so that specific areas of AI that cause harm, such as deepfake porn of nonconsenting individuals, are deterred and punished, while promoting an avenue for the utility of AI to thrive.⁷⁰

Cover Photo generated by DALLE·2 AI System

Jacob T. Metzger

Jacob Metzger is a 3L at the University of Cincinnati College of Law. Jacob received his bachelor's degree in Paralegal Studies at the University of Toledo with a focus in business and communications. Jacob is the editor-in-chief of the University of Cincinnati College of Law Intellectual Property and Computer Law Journal. He joined Law Review to write about consumer protection, intellectual property, advertising law, technology law, and topical topics.

References

1
Bergur Thormundsson, Global Total Corporate Artificial Intelligence (AI) Investment from 2015 to 2021, Statistica Rsch. Dep’t (June 23, 2022), https://www.statista.com/statistics/941137/ai-investment-and-funding-worldwide/.
2
Wex Definition Team, Artificial Intelligence, Legal Info. Inst. (Mar. 2022), https://www.law.cornell.edu/wex/artificial_intelligence_(ai).
3
Kevin Rose, An A.I.-Generated Picture Won an Art Prize. Artists Aren’t Happy., New York Times (Sept. 2, 2022), https://www.nytimes.com/2022/09/02/technology/ai-artificial-intelligence-artists.html; Bobby Allyn, A Robot Was Scheduled to Argue in Court, Then Came the Jail Threats, Nat’l Pub. Radio (Jan. 25, 2023), https://www.npr.org/2023/01/25/1151435033/a-robot-was-scheduled-to-argue-in-court-then-came-the-jail-threats; Pia Ceres, ChatGPT Is Coming for Classrooms. Don’t Panic, Wired (Jan. 26, 2023), https://www.wired.com/story/chatgpt-is-coming-for-classrooms-dont-panic/.
4
Vivian Lam, Human Art Already Has So Much in Common with AI, Wired (Feb. 24, 2023), https://www.wired.com/story/generative-art-algorithms-creativity/.
5
Dall-e, https://openai.com/dall-e-2/ (last visited Feb. 24, 2023).
6
Rose, supra note 3.
7
Lam, supra note 4.
8
Rose, supra note 3.
9
James Vincent, The Scary Truth about AI Copyright is Nobody Knows What Will Happen Next, The Verge (Nov. 15, 2022), https://www.theverge.com/23444685/generative-ai-copyright-infringement-legal-fair-use-training-data.
10
Jonathan Bailey, The Wave of AI Lawsuits Have Begun, Plagiarism Today (Jan. 17, 2023), https://www.plagiarismtoday.com/2023/01/17/the-wave-of-ai-lawsuits-have-begun/.
11
Vincent, supra note 9.
12
Rob Salkowitz, AI Is Coming For Commercial Art Jobs. Can It Be Stopped?, Forbes (Sept. 16, 2022), https://www.forbes.com/sites/robsalkowitz/2022/09/16/ai-is-coming-for-commercial-art-jobs-can-it-be-stopped/.
13
Vincent, supra note 9.
14
17 U.S.C. § 102(a).
15
See Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53, 57-59 (1884) (defining authors as “he to whom anything owes its origin; originator; maker; one who completes a work of science or literature”).
16
Richard Lawler, The US Copyright Office Says You Can’t Copyright Midjourney AI-generated Images, The Verge (Feb. 22, 2023), https://www.theverge.com/2023/2/22/23611278/midjourney-ai-copyright-office-kristina-kashtanova.
17
Vincent, supra note 9.
18
Emma Roth, DoNotPay Chickens Out on Its Courtroom AI Chatbot Stunt, The Verge (Jan. 25, 2023), https://www.theverge.com/2023/1/25/23571192/donotpay-robot-lawyer-courtroom.
19
Kamron Sanders, What Is Technology-Assisted Review? (Tar), The Nat’l L. Rev. (May 19, 2022), https://www.natlawreview.com/article/what-technology-assisted-review-tar.
20
Legalzoom Home Page, https://www.legalzoom.com/ (last visited Feb. 25, 2023).
21
Sanders, supra note 19.
22
Anthony Davis, The Future of Law Firms (and Lawyers) in the Age of Artificial Intelligence, Am. Bar Ass’n (Oct. 2, 2020), https://www.americanbar.org/groups/professional_responsibility/publications/professional_lawyer/27/1/the-future-law-firms-and-lawyers-the-age-artificial-intelligence/; Megan Cerullo, AI-Powered “Robot” Lawyer Won’t Argue in Court After Jail Threats, CBS News (Jan. 26, 2023), https://www.cbsnews.com/news/robot-lawyer-wont-argue-court-jail-threats-do-not-pay/.
23
Cerullo, supra note 22.
24
Id.
25
Id.
26
Id.
27
Ian Sample, What Are Deepfakes – And How Can You Spot Them? The Guardian (Jan. 13, 2020), https://www.theguardian.com/technology/2020/jan/13/what-are-deepfakes-and-how-can-you-spot-them; Dave Johnson, What is a Deepfake? Everything You Need to Know About the AI-Powered Fake Media, Bus. Insider, (Aug. 10, 2022), https://www.businessinsider.com/guides/tech/what-is-deepfake.
28
Id.
29
Deepfakenews, Episode 3, TikTok (Feb. 20, 2023), https://www.tiktok.com/t/ZTRnC387k/; Johnson, supra note 27.
30
Id.
31
Id.
32
Id.
33
Bianca Britton, They Appeared in Deepfake Porn Videos Without Their Consent. Few Laws Protect Them., NBC News (Feb. 14, 2023, 3:48 PM), https://www.nbcnews.com/tech/internet/deepfake-twitch-porn-atrioc-qtcinderella-maya-higa-pokimane-rcna69372.
34
Id.
35
Karen Ho, Deepfake Porn is Ruining Women’s Lives. Now the Law May Finally Ban It., MIT Tech. Rev. (Feb. 12, 2021), https://www.technologyreview.com/2021/02/12/1018222/deepfake-revenge-porn-coming-ban/.
36
Id.
37
Alexandra Arko, Nudify Me: The Legal Implications of AI-Generated Revenge Porn, JD Supra (Feb. 16, 2023), https://www.jdsupra.com/legalnews/nudify-me-the-legal-implications-of-ai-2348218/.
38
Ceres, supra note 3.
39
Id.
40
Id.
41
Andrew Holt, Legal AI-d to Your Service: Making Access to Justice a Reality, Vand. J. Ent. & Tech. L. (Feb. 4, 2023), https://www.vanderbilt.edu/jetlaw/2023/02/04/legal-ai-d-to-your-service-making-access-to-justice-a-reality/.
42
ChatGpt, How Information Theory Can Be Applied To Improve e-Discovery, JdSupra (Jan. 26, 2023), https://www.jdsupra.com/legalnews/how-information-theory-can-be-applied-5247810/.
43
Id.
44
Id.
45
Id.; Britton, supra note 33; Roth, supra note 18; Vincent, supra note 9.
46
Salkowitz, supra note 12.
47
Lam, supra note 4.
48
Vincent, supra note 9.
49
Lam, supra note 4.
50
Vincent, supra note 9.
51
Cerullo, supra note 22.
52
Roth, supra note 18.
53
Cerullo, supra note 22.
54
Salkowitz, supra note 12.
55
Britton, supra note 33.
56
Id.
57
Cal Civ. Code § 1708.86 (Deering 2023); Ga. Code Ann. § 16-11-90 (2023); Va. Code Ann. § 18.2-386.2 (2023); N.Y. Civ. Rights Law § 52-c (LexisNexis 2023).
58
Ga. Code Ann. § 16-11-90 (2023); Va. Code Ann. § 18.2-386.2 (2023).
59
Cal Civ. Code § 1708.86 (Deering 2023); N.Y. Civ. Rights Law § 52-c (LexisNexis 2023).
60
Britton, supra note 33.
61
Ho, supra note 35.
62
Holt, supra note 41.
63
See Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53, 57-59 (1884) (defining authors as “he to whom anything owes its origin; originator; maker; one who completes a work of science or literature”).
64
This response was generated by Chat GPT, a language model trained by OpenAI, on February 25, 2023. For more information about Chat GPT and OpenAI, please visit https://openai.com/.
65
Id.
66
Id.
67
Id.
68
Vincent, supra note 9.
69
Holt, supra note 41.
70
Id.; Britton, supra note 33.