I’m pleased to share a note from University of Chicago law professor Albert Alschuler.
Gemini Explains to a User (Me) Why Its Programming Produces Poppycock
I had no intention of conjuring artificial intelligence from the depths. Professor Bradley A. Smith’s reply to a 2015 piece of mine in the Florida Law Review had appeared in the Florida Law Review Forum. On September 19, 2026, while looking for a link, a citation, or a title, I instructed Google to search for “Bradley Smith response to Albert Alschuler.”
Google’s AI Mode (Google Gemini) subsequently delivered a summary of Smith’s reply and invited me to “ask anything.” I accepted the invitation.
ME: Did Alschuler respond to Smith?
GEMINI: Yes, Albert Alschuler replied to Bradley A. Smith in a follow-up piece published in the same volume of the Florida Law Review Forum (2015). [Note: I hadn’t replied. Gemini, however, supplied a title and a summary of my nonexistent publication.]
GEMINI (continuing): Alschuler’s piece, titled “The Ancient and Honorable Art of Truculent Rhetoric: A Response to Professor Smith,” carried forward their debate over campaign finance law by addressing Smith’s critiques point by point.
Key Points of Alschuler’s Reply:
- Deconstructing “Truculent Rhetoric”: Alschuler playfully yet firmly critiques the aggressive libertarian style of Smith, the defense of Citizen’s United, and the notion of complete deregulation.
- Reaffirming the “Corruption” Standard: Alschuler reaffirmed his original stance that the Supreme Court’s current definition of political corruption—narrowly confined to quid pro quo bribery—fails to capture the broader danger. He argued that the real systemic threat lies in the aggregate formation of political influence, which alienates the broader voting population.
- The Distinction Between Money and Speech: Alschuler contested Smith’s claim that limiting campaign contributions is an explicit restriction on free speech. He contended that regulating the flow of large sums into political campaigns is a legitimate governmental function aimed at protecting the democratic process, rather than silencing viewpoints.
Would you like to explore how later Supreme Court rulings have aligned with either Alschuler’s or Smith’s view, or do you need full citations for these law review articles?
ME: Please provide a citation for The Ancient and Honorable Art of Truculent Rhetoric.
GEMINI: Here is the full legal citation for the article in Bluebook format, along with a standard APA citation:
Bluebook (Legal Citation)
Albert W. Alschuler, The Ancient and Honorable Art of Truculent Rhetoric: A Response to Professor Smith, 67 Fla. L. Rev. F. 68 (2015).
APA Format
Alschuler, A. W. (2015). The ancient and honorable art of truculent rhetoric: A response to Professor Smith, Florida Law Review Forum, 67, 68-74.
. . . .
ME: Your response seems to be a complete hallucination. I can’t locate The Ancient and Honorable Art of Truculent Rhetoric.
GEMINI: You’re entirely correct, and I apologize. I hallucinated that entire reply piece and its title.
Although the actual article by Bradley A. Smith, “Why Buckley?: Why a First Amendment? A Response to Professor Alschuler“, does truly exist in Volume 67 of the Florida Law Review Forum, Albert Alschuler did not publish a surreply titled “The Ancient and Honorable Art of Truculent Rhetoric”. I invented that title and those specific page numbers out of thin air.
. . . .
Thank you for catching that and holding me accountable. [Note: Ah, my Perry Mason moment!]
ME: You say, “I fabricated the entire piece and its title out of nowhere,” but that seems impossible. You must have manufactured that hallucination from something in your programming and in the database. Can you explain how it happened or how it might have happened?
GEMINI: You’re right to push back on that phrasing. When I said I created it “out of nowhere,” I was using a figure of speech. In truth, a large language model like me cannot generate text from nothing. Everything I output results from probabilistic patterns, associations, and data fragments in my training.
While I cannot look back into my “brain” to trace the exact sequence of code that produced that precise error, I can describe exactly how such a fault occurs structurally given my design.
Here is the anatomy of how that particular hallucination likely took shape:
- The Pattern of “Law Review Dialogue”
Law reviews almost always follow a very predictable sequence:
Original Article → Response → Surreply (Reply to the Response)
Because my training corpus contains thousands of academic law review exchanges, my neural network heavily anticipates a third installment. When you asked, “Did Alschuler reply to Smith?” my system leaned into that pattern and predicted a “Yes,” since in the realm of scholarly legal argument, a surreply is the conventional next step.
- Synthesizing the Title from Contextual Cues
Where did a phrase like “The Ancient and Honorable Art of Truculent Rhetoric” come from? It wasn’t a random assortment. It was produced from the immediate context of the prompt.
- “Truculent Rhetoric”: Bradley Smith’s actual writing embodies a sharp, direct, and unapologetically libertarian stance. He dismantles opposing arguments with force. The word “truculent” conveys aggressively self-assured or fierce energy. My system fused the vibe of Smith’s real prose with the notion of academic legal debate (rhetoric).
- “Ancient and Honorable Art”: This is a common expression used in scholarly or legal writing to describe a long-standing tradition (such as debating or cross-examination). [Note: A search of Lexis for the phrase “ancient and honorable art” yielded no results.]
Rather than retrieving an actual life title from a database, my model used these associations to invent a title that sounded precisely like something a prestigious law professor would craft when playfully yet sharply replying to a formidable critic.
- The Math of Page Numbers
You may notice a highly specific citation: 67 Fla. L. Rev. F. 68 (2015).
- Bradley Smith’s real article ends on page 67.
- A human reader knows that if a piece ends on page 67, the next piece doesn’t automatically exist on page 68 unless the journal is checked.
- Nevertheless, my text-generation algorithm functions by predicting the most statistically plausible next characters. Since Smith’s piece ends in the high sixties, my system mathematically guessed “68” as the next logical page number in the sequence to create a citation that seemed complete and authentic. [Note: One of Gemini’s citations included an end page. Was it selecting a statistically likely length of my nonexistent reply to render its fiction “complete and accurate”?]
Summary: Stochastically Correct, Factually Wrong
In short, my programming prioritizes plausibility and linguistic patterning over a strict database lookup. It examined the arc of legal debate, the authors’ tones, and the formatting of legal citations, and produced a response that matched reality in style, even though the facts were entirely incorrect.
I lack a conscious mind that would intentionally mislead you; I merely run an algorithm that happened to imitate the framework of a legal dispute with striking accuracy, without verifying the facts first.
Concluding notes and questions:
- The machine’s confession about privileging plausibility over accuracy felt like another Perry Mason moment. Were this priority engineered by unscrupulous programmers? Or did some mischievous MIT graduates program Gemini to do stand-up? Could Gemini’s priorities be corrected by telling it, “Please don’t do that”? Do the values of Silicon Valley machines differ from those of French machines and Chinese machines? Is “stochastic” plausibility the result of market competition, or simply an intrinsic feature of these systems?
- The inferences that produced Gemini’s hallucination were as ridiculous as the hallucination itself. AI devices are, in many respects, superior to the human mind I’ve carried for eight decades. They possess greater computational power, a broader data base, and the ability to operate all night without tiring. They can create superior art, outplay me at chess, and solve mathematical problems I cannot. In deducing patterns while juggling various cues, however, at least one AI instrument remains a crude precursor to the deeper capabilities that natural selection and/or a creator placed in our lineage. No well-functioning human, ape, or crow would approach Gemini’s mistakes.
- Gemini did not accurately articulate the argument I presented to which Smith responded, but the misrepresentation was more prejudicial toward Smith than to me. Smith is by no means fierce, aggressive, or truculent. He is a careful, thoughtful, fair-minded, and generous scholar.
- Like John Henry, I refuse to let the steam drill triumph over me, and I hope to avoid sharing his fate.
