What One AI Commercial Reveals About Accuracy, Expertise, and Trust
A predictive market firm used AI to make "the best AI commercial ever." It got Paul Revere wrong, the Wright brothers wrong, and the Battle of Thermopylae wrong. It was, in fact, perfect — just not in the way they intended.
By R.R. Hicks · Forlorn Hope Publishing
When AI Gets the Details Wrong
'Epic historical footage is perfect for AI... but if you're going to send it out to the world, it would help to have someone who, I don't know, maybe has some historical knowledge riding shotgun at the old AI console.
Kalshi — a predictive market firm that claims to have called the 2024 election when no one else did — released what it called, and I quote:
KALSHI'S OWN DESCRIPTION
"This may be the best AI commercial we've ever made. We worked with one of the best directors in the world to create a wild new ad for Kalshi (Last one hit 100M views). Here's a breakdown of how we made it in ONE WEEK. Everything starts with the script. The Kalshi team asked us to think of a way to showcase their brand in a way that really highlights underdogs throughout history. Epic historical footage is perfect for AI, so it seemed like a great angle to take for this video."
A Few Things About "Epic Historical Footage"
- Paul Revere (a) did not shout, "The British are coming" to people who identified as British, (b) he rode through a quiet countryside — not a burning, bombarded inferno of a town — which, by the way, made his warning somewhat superfluous.
- The Wright brothers' first flight wasn't made from a wheatfield, and they didn't start with a 1917 Sopwith Camel–type aircraft that they could only have dreamed of.
- Leonidas and the Spartans didn't attack the Persians—if they had, they probably wouldn't be remembered at all, and the world would have had to live without Gerard Butler's 300.
"If you can't get Paul Revere right, why would I trust you to get the next presidential election?"
The Punchline Writes Itself
Here's the punchline: Kalshi is in the history business. Their entire model is about using data, judgment, and "expertise" to predict what happens next. They claim to have nailed the 2024 election when no one else did. But if this is the level of historical rigor in the ad, how is anyone supposed to reconcile giving them real money, let alone trusting the odds?
If you can't get Paul Revere right, why would I trust you to get, say, the next presidential election, or even a Thursday night football game?
It's one thing for AI to spit out "epic" visuals. But in a business that's supposed to be about knowledge — about knowing the difference between actual history and something you saw in a Michael Bay movie — maybe having an informed human at the controls isn't just a good idea, it's essential.
The Broader Principle
What happened to Kalshi isn't a cautionary tale about AI specifically. It's a cautionary tale about confidence without expertise — the Jerry Rosenberg problem applied to marketing. Supremely confident, technically capable within a narrow band, and wrong in a way only someone who actually knows the subject would catch.
For law firms, the stakes are different in kind but not in structure. The question isn't whether AI can produce content that looks right. It can. The question is whether anyone with expertise is reading it before it goes out. And whether "one week" is actually enough time to know the difference between Thermopylae and a Michael Bay set piece.
We Know Who Was at Thermopylae
Creative Legal Writing puts experienced writers and lawyers — not a Sopwith Camel — at the controls. Content that's accurate, argued, and built from real knowledge of what happened and why it matters.
Looking Right Isn't the Same as Being Right
An AI can paint Paul Revere against a dramatic sky. It takes someone who's actually read the history to know why the scene doesn't work.
Legal content isn't much different.
The words can be polished. The structure can be clean. But if no one at the keyboard knows the subject, confidence starts masquerading as expertise.
That's not a trade we'd make.
If your firm is looking for legal content written by people who care as much about getting the facts right as getting the writing right, schedule a conversation.
Frequently Asked Questions
Yes. AI can produce convincing text, images, and videos that contain factual, historical, or contextual errors. Human review is essential to verify accuracy before publishing content, especially in industries where credibility and expertise are important.
Yes. AI can summarize information quickly, but it may introduce inaccuracies, omit important context, or present incorrect information with confidence. Fact-checking by someone familiar with the subject helps reduce these risks.
Legal marketing helps establish trust with potential clients. Even small factual or legal inaccuracies can damage credibility and create confusion about a firm's knowledge and professionalism.
AI can assist with drafting and research, but it cannot replace the judgment and expertise of professionals who understand the nuances of a particular field. Expert review remains an important part of producing reliable content.
