By: Simonne Kapadia
I. Defining artificial intelligence
The first step in considering the ethical use of artificial intelligence is to understand what artificial intelligence is. There are two broad categories of artificial intelligence – predictive and generative. Predictive artificial intelligence is “a computer program that uses statistical analysis to predict future events, anticipate behaviors, and identify patterns.”[1] We’ve all encountered predictive artificial intelligence – whether it’s our favorite streaming service recommending a new song we may like or an online retailer recommending our next potential purchase.[2]
Generative artificial intelligence, conversely, uses reference databases to create or generate content, like images or text, in response to a submitted prompt.[3] Generative AI became part of the popular discourse in late 2022 with the release of OpenAI’s ChatGPT.[4] A submitted prompt, in the context of ChatGPT, can be a simple question – like what’s the weather like in a particular part of the country during a particular time of year.
II. Regulating artificial intelligence
Using artificial intelligence ethically means, at a minimum, complying with existing AI-regulations. So, now that we have a better understanding of what artificial intelligence is, let’s talk about how the legislature and other authoritative bodies are regulating it.
One of the main concerns with using AI involves informing consumers of AI-generated content that it is, in fact, AI-generated. Recently enacted and proposed rules and regulations demonstrate this concern. In February 2024, Michigan amended its state election laws to regulate the use of AI. This amendment requires persons, as defined in the statute, to disclose when content, like a prerecorded telephone message or political advertisement, is “generated in whole or substantially by artificial intelligence.”[5] This amendment defines artificial intelligence as:
a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments, and that uses machine and human-based inputs to do all of the following:
- Perceive real and virtual environments.
- Abstract such perceptions into models through analysis in an automated manner.
- Use model inference to formulate options for information or action. [[6]]
Notably, this definition of artificial intelligence mirrors that used in federal legislation.[7]
The Michigan legislature is also attempting to regulate AI-generated content outside of the election context. In June 2024, the Michigan House passed House Bills 5569 and 5570, which propose civil and criminal penalties for producing and distributing nonconsensual, sexually explicit “deep fake” content. House Bill 5569, known as the “protection from intimate deep fakes act” defines “deep fake” as:
a video recording, motion-picture film, sound recording, electronic image, or photograph, or a technological representation of speech or conduct substantially derivative of such a recording, film, image, or photograph, to which both of the following apply:
- It is so realistic that a reasonable person would believe it depicts speech or conduct of a depicted individual.
- The production of it was substantially dependent on technical means, rather than the ability of another individual to physically or verbally impersonate the depicted individual.[8]
So, these bills attempt to penalize the creation and distribution of sexually explicit AI-generated fake content that, to a reasonable person, looks real.
The federal government and federal courts are also attempting to regulate artificial intelligence so that the audience of the content is aware that the content is AI-generated. In October 2023, President Biden issued an executive order that, among other things, requires government agencies to determine how to label “synthetic” content, such as by watermarking this content.[9] The executive order defines “synthetic content” as “information, such as images, videos, audio clips, and text, that has been significantly modified or generated by algorithms, including by AI.”[10]
In January 2024, the United States District Court for the Eastern District of Michigan proposed a local rule requiring filers to disclose if they used artificial intelligence to generate their filing.[11] This proposed regulation further requires filers to attest that a human verified the citations in the filing and checked the citations for accuracy.[12] Notably, however, the United States District Court for the Western District of Michigan does not have a similar proposed local rule or enacted local rule. Neither does the United States Court of Appeals for the Sixth Circuit. Further, in June 2024 the United States Court of Appeals for the Fifth Circuit decided against adopting a proposed rule requiring filers to disclose their use of artificial intelligence.[13]
Importantly, it’s not only new rules and regulations that may impact an individual’s use of artificial intelligence. For example, existing rules of professional responsibility necessarily impact an attorney’s use of artificial intelligence (as with any other form of technology).[14] And in February 2024, the Federal Communications Commission confirmed that existing Telephone Consumer Protection Act regulations extend to AI calls that generate human voices.[15]
The sphere of artificial intelligence regulation is an evolving space. Applying new and existing regulations to this technology will continue to change as the technology itself changes.
[1] TED AI San Francisco, Glossary: Predictive Artificial Intelligence, TED AI https://tedai-sanfrancisco.ted.com/glossary/predictive-artificial-intelligence/ (emphasis added) (all websites last accessed July 15, 2024).
[2] Coursera Staff, Generative AI vs. Predictive AI: What’s the Difference?, Coursera (June 5, 2024), https://www.coursera.org/articles/generative-ai-vs-predictive-ai; Jennifer Monahan, Artificial Intelligence, Explained, Carnegie Mellon University Heinz College, https://www.heinz.cmu.edu/media/2023/July/artificial-intelligence-explained.
[3] Generative artificial intelligence, Merriam-Webster, https://www.merriam-webster.com/dictionary/generative%20artificial%20intelligence.
[4] Bernard Marr, A Short History of ChatGPT: How We Got To Where We Are Today, Forbes (May 19, 2023, 1:14AM), https://www.forbes.com/sites/bernardmarr/2023/05/19/a-short-history-of-chatgpt-how-we-got-to-where-we-are-today/. GPT stands for “generative pre-trained transformer.” Id.
[5] MCL 169.247(7); MCL 169.259(1)(a).
[6] MCL 169.202(1).
[7] 15 USC 9401(3).
[8] 2024 HB 5569.
[9] Executive Order No. 14110, Sect. 4.5(a)(ii).
[10] Id. at Sect. 3(ee).
[11] Proposed LR 5.1.
[12] Id.
[13] Compare Proposed LR 32.2 with Fifth Circuit Decision on Proposed Rule.
[14] See, e.g., State Bar of Michigan, Artificial Intelligence for Attorneys – Frequently Asked Questions, https://www.michbar.org/opinions/ethics/AIFAQs.
[15] FCC Declaratory Ruling, 24-17.
