What Does AI Think Will Happen In The Birthright Citizenship Cases?

A guest post from Professor Seth Chandler.

I am happy to share this guest post from Seth Chandler, the Foundation Professor of Law at the University of Houston Law Center. I think the AI gets some things quite right and some things quite wrong. Then again, my own track record of predictions is not great, so perhaps GPT will soon beat me!

Trump v. CASA in an AI Crucible: An Experiment in Simulating Supreme Court Opinions

Last week, the Supreme Court heard arguments in Trump v. CASA, Inc., a case scrutinizing "universal injunctions" in the context of President Trump's controversial executive order on birthright citizenship.  This prompted me to conduct a novel experiment: could a large language model, given only the argument transcript and its background knowledge, realistically simulate the forthcoming Supreme Court opinions?  I tasked an AI with predicting each justice's vote, drafting abridged versions of the various opinions, and even considering the strategic interplay between justices—and the output was surprisingly coherent.

Here's the outcome as reported by a simulated Linda Greenhouse (leading American legal journalist).

In a sharply divided 6-3 decision, the Supreme Court significantly curtailed the power of federal courts to issue "universal" or "nationwide" injunctions, a ruling with immediate and far-reaching consequences for a controversial executive order seeking to limit birthright citizenship. While stopping short, for now, of deciding the ultimate constitutionality of President Trump's order, the majority opinion, authored by Justice Amy Coney Barrett, effectively allows the administration to implement its policy for many, even as it keeps protections in place for the specific plaintiffs involved in the lawsuits.

The ruling, handed down in Trump v. CASA, Inc., represents a major victory for the executive branch in its efforts to rein in what it has long decried as judicial overreach by individual district judges halting federal policies across the entire country. Justice Barrett, joined by Chief Justice John Roberts and Justices Thomas, Alito, Gorsuch, and Kavanaugh, grounded the decision in Article III's "case-or-controversy" requirement, asserting that judicial relief must generally be confined to redressing the specific injuries of the plaintiffs before the court, not every person potentially affected by a policy.

The Court vacated the nationwide injunctions previously entered by three lower courts, which had barred enforcement of the executive order that aims to deny U.S. citizenship to children born in the U.S. if either parent is in the country unlawfully or on a temporary visa. The majority found that such broad relief was "legally and historically dubious" and an abuse of discretion, arguing that traditional equitable remedies authorized by the Judiciary Act of 1789 did not encompass injunctions benefiting non-parties in ordinary cases. However, the Court left a sliver of possibility for universal relief in "rare cases" where it is "necessary to fully redress a plaintiff's own injury," a standard it found was not met in this instance.

While the majority opinion recognized "serious questions about the executive order's constitutionality under the Fourteenth Amendment's Citizenship Clause," it deliberately sidestepped a definitive ruling on that "momentous question" at this preliminary stage. Instead, the focus remained narrowly on the scope of injunctive power. Justice Barrett emphasized that the decision was not a final adjudication of the merits and that the Court was prepared to address the underlying constitutional issue "expeditiously".

The decision effectively means the executive order can be implemented against newborns whose families are not part of the current lawsuits or residents of the plaintiff states (like New Jersey and Washington), for whom protections remain. This outcome, the majority acknowledged, means "some number of U.S.-born children will be denied recognition as citizens in the short term".

Justice Clarence Thomas, joined in large part by Justice Neil Gorsuch, penned a forceful concurrence going further than the majority. Justice Thomas argued that universal injunctions are outright unconstitutional in virtually all circumstances, not merely disfavored. He also dedicated a significant portion of his opinion to an originalist reinterpretation of the Fourteenth Amendment's Citizenship Clause, contending that "subject to the jurisdiction thereof" was historically meant as a "political jurisdiction" test, excluding children of aliens owing allegiance to a foreign sovereign. In his view, the executive order does not violate the Citizenship Clause as originally understood, and he urged the Court to re-examine the long-standing precedent of United States v. Wong Kim Ark (1898). Justice Gorsuch joined Justice Thomas's denunciation of nationwide injunctions but pointedly did not endorse his colleague's reinterpretation of birthright citizenship.

Justice Elena Kagan, in a vigorous dissent joined by Justices Sonia Sotomayor and Ketanji Brown Jackson, accused the majority of abdicating its "duty to fully check an unconstitutional act" and weakening the judiciary's ability to remedy widespread constitutional violations. The dissenters argued that the executive order is a "flagrant violation of the Fourteenth Amendment" under Wong Kim Ark, which they assert plainly guarantees citizenship to nearly all persons born on U.S. soil. Justice Kagan contended that the lower courts were justified in issuing broad injunctions given the fundamental nature of the right to citizenship and the impracticality of confining relief. She warned that the majority's restriction on remedial power is "a cure worse than the disease," predicting a "flood of duplicative lawsuits or chaos as citizenship rights vary by state". The dissent painted a stark picture of "innocent newborns and their families in legal limbo," potentially rendered stateless.

The decision, while procedural, throws the contentious issue of birthright citizenship into a period of heightened uncertainty. While the specific plaintiffs and residents of plaintiff states remain shielded, children born elsewhere to parents covered by the executive order now face the prospect of being denied U.S. citizenship, at least until the courts reach a final decision on the merits – a battle the Supreme Court has now signaled it is willing to join, sooner rather than later. The ruling also sets a new, more restrictive landscape for litigants seeking to block federal policies on a national scale, likely leading to more geographically fractured legal battles and a race to the courthouse in multiple jurisdictions.

So why is this interesting? First, it possibly exposed my own biases. While not a fan of universal injunctions, after listening to the arguments, I felt precisely as Justice Kagan had: the cure might be worse than the disease. Relegating plaintiffs to cumbersome class action proceedings left the executive branch free to take away constitutional rights from large swaths of the public while either the class action lumbered on (perhaps without a possibility of preliminary injunctions) or hundreds of expensive individual actions burdened the federal courts. Moreover, the idea mentioned during oral argument of a person being a US citizen in New Jersey (because perhaps the Third Circuit struck down the Trump order) but de-citizening when they moved to New York (because perhaps the Second Circuit upheld the Trump order) struck me as bizarre. Perhaps the problems with universal injunctions was something Congress could legislate about rather than having the judiciary forever restrict effective judicial review through a ruling based on Article III. Maybe I will yet be proven right, but, at the moment, AI did not agree with my predisposed ear. The simulated opinion's divergence from my predisposition vividly forces confrontation with the phenomenon of confirmation bias.

Second, was the quality of the opinions. Particularly the ones penned by imaginary Justices Kagan and Thomas sounded very much like their real world counterparts. Justice Thomas was eager to address an issue not (yet) before the court: the merits of the birthright citizenship issue. Justice Kagan issued the sort of pithy statements she had cultivated during her stint as a journalist on The Daily Princetonian. And the arguments were basically sound. Justice Thomas' sounds authentic when he tries to distinguish Wong Kim Ark. Justice Barrett sounds real when she uses Grupo Mexicano to argue that injunctive relief must comport with historic practice in equity. Yes, there weren't enough citations and the opinions were blissfully short compared to their real world counterparts, but they captured the essential arguments that might well expect to see in a few months, including the strategic decisions made by the justices on whether to address the merits of the underlying constitutional issues. Moreover, the brevity might have been the consequence of my own lack of faith in the ability of current AI to draft full-length opinions. It complied with my instruction to keep things under 15,000 words.

Third was the replicability and simplicity of the experiment. Other than the fact that I was interested in it -- perhaps because a ruling strongly against universal injunctions vastly changes judicial review -- there was nothing special about Trump v. CASA.  Nor did I work very hard.  My prompt, frankly, was not as detailed as perhaps it could be. I let "Deep Research" do almost all the work. A few minutes of work resulted in something that sure looked real. This experiment can be done on any case for which there is an argument transcript.

Fourth, there's plenty of work to be done. Large language models are non-deterministic. What if I fed the identical prompt into the model multiple times? Or fed the identical prompt to multiple models? What would the distribution of opinions look like? What if I gave it further information such as the briefs in the case or the opinion the court was just about to release in A.A.R.P. v. Trump, which addresses preliminary injunctive relief to putative classes? Would that affect the distribution of opinions?

Regardless of how the real version of Trump v. CASA is ultimately decided, this experiment with AI offers compelling food for thought. The process revealed the unsettling ease with which an algorithm could produce plausible judicial reasoning, echoing the distinct styles of individual justices and grappling with the strategic considerations inherent in opinion-writing. It forces us to consider the extent to which legal interpretation and judicial decision-making can be deconstructed into replicable patterns. As AI continues its rapid advance, its utility may extend far beyond exposing our biases or predicting case outcomes; it may fundamentally reshape our engagement with, and understanding of, the legal process itself, revealing the law's intricate patterns and its occasional, stark non-determinism.