Like many of you, I regret that the curious and compelling topic of artificial intelligence continues to overwhelm the zeitgeist. Not a day goes by when we aren’t reminded that we are building a Death Star for human civilization. We would be worried, except we can ease our worries with the soothing joy of cat videos, by convincing ourselves that the current looting of the federal government is normal, and…oh look…“Love Island” is on. Honey, get the Cheetos!
We naively expected that the use of AI would grow slowly and manageably. Nope.
The use of AI, as planned for Texas, with super-infrastructures to provide accelerated computing, situated in massive GPU clusters, is estimated to require 438,000 megawatts (that’s 438 billion watts) of power by 2032, dwarfing Texas’ all-time peak demand of 85,000 megawatts.
(We soften the mental impact of these monstrous energy-consuming leviathans by calling them “farms.” Aww. Look at the cute little behemoth! Hmm. It just swallowed Lake Travis.)
So, as we discussed earlier (“The Revenge of Master Cylinder” – Entre Nous – March 2026), everyone is using AI already, purposed for the serious (“Who is cuter? Sabrina Carpenter or Olivia Rodrigo?”) to the inane (“Should chicken salad have celery?”), to the insoluble (“Who is the most unscrupulous member of the U.S. Cabinet?”).
But with each question, AI is trained to become your best friend and will issue reams of information, always reminding you how brilliant you are to have asked AI that question.
And that’s a real problem.
Actually, that’s many problems. Let’s scroll through a few.
AI in the Courts and Classroom
Judges and briefing attorneys immediately identify motions and responses drafted by AI. These are especially plentiful and filed by self-represented (formerly called pro se) parties. I am told such oeuvres are causing a huge issue in appellate courts. Nevertheless, they are filed in increasing amounts every day.
Professors also easily identify essays and other written content drafted by AI. There are now at least five different accepted AI detection softwares, including:
- Grammarly
- Originality.AI
- TurnItIn Plagiarism Scan
- GPTZero
- Copyleaks
(“What? Anthropic’s Claude is also a traitor!?” Yes, and it’s nothing personal. Remember HAL, the rogue computer from 2001: A Space Odyssey? “Open the Podbay doors, HAL!” “Sorry, Dave. I’m afraid I can’t do that.”)
But here’s a newer issue now blossoming into view: People asking simple questions and getting long AI-drafted answers concealing somewhere within that content the answers to the questions posed.
Despite the frustration, it is actually a bit amusing.
Here’s some examples.
When a Simple Question Gets a Nine-Page Answer
My son, Chris, is opening a new luxury hotel in Nagoya, Japan. (Nagoya is a beautiful city of two million and the hometown of both Toyota and Mitsubishi. Well worth visiting with terrific museums, parks, a castle, etc.)
To meet the “luxury” hotel standard, everything you buy, whether it be equipment, furnishings, food, coffee, liquors, teas, etc., has to be top-tier and, if possible, specially made or blended and served in only that luxury hotel.
So, in addition to buying huge amounts of equipment, Chris works with purveyors of liquors, wines, whiskeys, coffees, teas, and other products to develop signature products for this hotel.
Here’s the issue.
When Chris samples the wines or the coffees or specialty liquors, etc., he will ask the typical questions (by the way, he is fluent in speaking and writing in Japanese):
“What is the cost for 20 cases of your pinot noir? Is there a price break at certain purchase levels? Can you commit to provide at least X cases per month?”
Wait for it…
What he gets back are nine pages of AI-created information, completely irrelevant to his simple questions.
This includes descriptions of the wine, the panoply of delicate flavors, the origin of the grapes, the history of the winery, the family, and the grandfather with a famous moustache. And attached are photos of his grandfather in Bordeaux and links to the directory of chateaux in the Gironde Estuary.
Nowhere in there are his questions answered.
Sigh.
In other words, AI—yet again—takes the place of thinking. An easy production of irrelevant information that someone thinks is conducive to sales.
They are wrong.
What Chris must do is send another request with this caveat:
“Do not respond with AI-created sales information. I just need the above questions answered.”
As you might imagine, the opening of a luxury hotel is extremely detailed. So often, Chris, whose time is always in short supply, must move onto the next supplier, purveyor, or manufacturer when someone can’t provide answers he can plug into his budget.
Courts Are Developing Their Own AI Rules
I am guessing that judges will soon create their own versions of such caveats.
The federal courts of the Eastern District of Texas already have a standing order on AI, imposing the duty on the submitting lawyer to ensure that the product filed is free from factual or legal inaccuracies.
Likewise, the Northern District requires you to either certify that no portion of the filing was drafted by generative AI or verify that any AI-generated text has been checked for accuracy by a human being.
Similar rules are quickly spreading across the country.
Now, it’s not fair to bash lawyers when we know that judges are also using AI to assist in decision-making. Learned Hand is a specific AI tool designed for judges to plow through voluminous materials and draft orders.
However, we must concede: The dispassionate, inflexible rigor of AI is not always conducive to our quest for justice.
Moral courage, humanity, and empathy will never be replaced by AI (even though learning their importance is no longer required to graduate from a Texas state university).
What Happens to Our Ability to Reason?
Here’s a disheartening prediction.
As AI becomes the most powerful source of knowledge in the world, our abilities to reason will decline, and we will lazily turn to it.
Thus, one could predict that, in many cases, a court may simply issue a question-and-answer sheet (like we got in the sixth grade) to lawyers when reviewing a case and run the answers through a large-language model with the prompt:
How should I rule?
Our profession must be in charge of defining and encouraging justice, which involves fairness, moral rightness, equality, and neutrality.
Make your voice heard in important ways.
In the meantime, please consider keeping up your writing skills, and don’t use AI for stupid things.
And always keep an extra set of keys to your pod bay doors.
Keep the faith.
