National AI legislation would create much-needed fairness and equity in 2024
By: Andreu Beltran
Most would agree that generative artificial intelligence has introduced fascinating new opportunities. However, it is also a legal gray zone because the topic raises a multitude of ethical and legal questions: is AI protected by copyright law? How can racial and economic bias in AI be prevented? Who is liable if AI causes harm? While many states have already established AI legislation, the federal government must set a national standard to make future AI decisions more equitable and peaceful on topics ranging from the hiring process to the 2024 election.
The first issue that must be addressed is discrimination resulting from AI use. If an AI is choosing between two equally qualified candidates for a position, it unfairly judges candidates based on qualities like race, gender, or sexual orientation. This can either be due to the biases of the programmer or simple carelessness in the code. To prevent prejudice from arising by accident in AI, developers must use data augmentation or introduce AI to a variety of perspectives in training. However, federal legislation is required to make this mandatory. Otherwise, each state could use AI differently, making it more likely that minority residents of one state will be at a disadvantage compared to those in other states. Only a few states have restricted AI in hiring practices, such as New York, California, Illinois, Maryland, Massachusetts, and New Jersey—there needs to be more. AI discrimination is also relevant in insurance as companies use predictive modeling to determine a client’s insurance plan. However, Colorado is the only state that requires insurance companies to disclose the use of AI, meaning that clients in other states may be facing discrimination by AI without knowing it. To maximize economic equity across the nation, AI legislation must be uniform. Otherwise, prospective employees and customers in AI-free states will be treated differently than those in AI states.
With the 2024 presidential election primaries having already begun, more AI legislation is necessary to make the campaign trail truthful and just. The main threat with AI in this regard is deepfakes, which digitally manipulate one’s face to look like someone else’s. This can be done with cheap technology, making it easy to impersonate a presidential candidate and release it to the internet, potentially swaying voters. Although several states, like Washington, Texas, Minnesota, Michigan, and California, all have legislation to prevent deepfaking, without uniform policy nationwide around impersonation, misinformation will undoubtedly spread online, and it will become difficult to determine how genuine a candidate may be. Misinformation with AI is particularly dangerous because it is much more difficult to confirm whether or not it is legitimate: a fake Trump tweet can easily be disproved by checking social media, but a fake video or audio recording of him would be more difficult to disprove. However, that is just one harm of deepfakes. The technology can create a lot of animosity and confusion among voters. Given the controversy around the past presidential election, keeping falsehood out of this upcoming election is particularly important: 57% of Republicans still think that Joe Biden was not elected legitimately. While politicians’ beliefs heavily sway the voter opinion, if the United States does not prevent vindictive campaigning through deepfaking, truth and fiction will become muddled, and there will be no way of proving that the election was legitimate. If enough violence is created that United States citizens do not feel as if their rights are protected, another insurrection like that of January 6th, 2021, is completely possible. However, deepfake technology is not just a threat to presidential candidates. Any public figure is in danger of being misrepresented in the media. Therefore, the federal government needs to quickly create policies that expel deepfake technology from social media and prevent it from spreading lies.
Discrimination and misinformation are just scratching the surface of AI’s legal issues. However, retaining different policies for each state will inevitably create more inequity in the long run, including in the 2024 election, future job hiring, or even the college admissions process. Therefore, to counter these potential dangers, it is imperative that the federal government passes more AI legislation.