Case study
Using AI to Sell a Game Show
Matt O’Brien, October 2026
Long before the advent of Artificial Intelligence, I heard a fascinating (and slightly terrifying) story about how Target somehow knew a 16–year-old girl was pregnant before her own father did. Little did I know, these same types of data insights would eventually spawn two seasons of a primetime gameshow called The Algorithm … with a US version currently in the works with Steven Spielberg’s Amblin TV.
The Algorithm (TV3 Ireland)
Back to Target. Using data analytics, Target looked at their baby-shower registry to observe how shopping habits change as a due date approaches. The company saw that during the first 20 weeks, pregnant women often stock up on supplements like calcium, magnesium and zinc. Using these insights and others, they created a “pregnancy-prediction model” identifying tens of thousands of women across America who were likely expecting.
About a year into the program, an angry father stormed into his local Target, clutching a pile of coupons for baby clothes and cribs sent to his teenage daughter. The manager of the store apologized profusely for the mix-up and called the father a few days later to follow-up. In a surprise twist, the father had since discovered his daughter was in a secret relationship … and was expecting a baby in August.
How could a giant corporation know something so intimate about someone’s personal life? We all know that Facebook, Google and others have powerful prediction algorithms designed to reduce each user down to a series of 1’s and 0’s. This led me to ask a simple question that would end up becoming the central premise of the show:
Can a computer know you better than your own family?
I already had a basic idea of the format.
1. Put Mom in a soundproof booth with her husband, family and friends in studio.
2. Ask Mom a subjective question. Mom secretly locks in her answer.
“What turns you off the most?”
a) Bad Handshake
b) Bad Listener
c) Bad Hygiene
3. Show The Algorithm’s prediction and reasoning to viewers at home.
“The Algorithm predicts Mom will choose ‘Bad Listener’ because [insert data driven reason].”
4. Reveal Family’s answer.
“We chose Bad Hygiene.”
5. Reveal Mom’s Answer.
“I chose Bad Hygiene.”
6. If Family is right, they move up the money ladder. If Family is wrong and Algorithm is correct, they lose.

My original handwritten notes for the basic “flow” of the game.
But this was ten years ago and in 2013, no one really knew what an “algorithm” was. It’s one thing to hear how powerful computers are at predicting our behavior … it’s another thing to experience it first-hand. To have a chance of selling this show, I’d have to “show” and not “tell” buyers exactly how the Algorithm knew them better than their friends/family/co-workers.
To do that, I needed to feed the Algorithm data.
My first stop was SurveyMonkey. I created a list of 100 sample questions. Things like:
“You’ve just found out you’re having a baby. Which would you prefer?
a. Baby boy
b. Baby girl
c. Identical twins”
“Neighborhood kids spray paint your garage with three bad words and you only have enough paint to cover up two of them. Which one do you leave?
a. The N word
b. The C word
c. The F word
“Which animal best describes you in the bedroom?
a. Lion
b. Horse
c. Cobra
I blasted it out to as many people as I could on SurveyMonkey. Participants answered a set of demographic questions (e.g. Age, Gender, Education, Birth Sign, Income, etc.). Once I had a enough responses, it was time to sift through the data to find trends/correlations. To do that, I found a nifty formula called a Naïve Bayes classifier:

Naïve Bayes classifier
Turns out, this is the basic algorithm behind Spam filters. If you get an e-mail that has the words “grow” “bigger” and “pills,” there’s a 99.99% chance it’s spam. Using that same approach, we can use the Naïve Bayes classifier to identify correlations in the questions and answers. For example, in our earlier question:
“What turns you off the most?”
a) Bad Handshake
b) Bad Listener
c) Bad Hygiene
85% of respondents from the UK replied “Bad Hygiene” to this question.
77% of respondents from the US replied “Bad Handshake” to this question.
That meant if I was pitching commissioners at the British Broadcasting Corporation (BBC) in London, the Algorithm would predict “Bad Hygiene,” but if we were on the FOX lot in Culver City, CA, the Algorithm would predict “Bad Handshake.” Like a gambler at a blackjack table, the Algorithm simply places a bet on the answer that was most likely given the player’s demographic info.
As I went through the data, I discovered other patterns that I couldn’t quite explain, but I didn’t need to. The Algorithm just needed to get the answers correct based on whoever was playing the game.
“Which shade of green do you prefer?”

Fun fact: it turns out that when women see three shades of the same color, over 80% of the time, they will choose the “middle” shade as their favorite. With men, the results were more or less split across the three choices. Again, I didn’t need to know “why” this was happening … just that if the Algorithm is making a guess on this question and the player is a woman, it’s going to choose the middle shade. In other words, the odds are always in the Algorithm’s favor.
With a crude dataset in hand, it was time to figure out how to bring this data to life. The holy grail in any game show pitch is to have the executives play the game. Could we show that The Algorithm knows the Executive better than their co-workers?
I wanted to have the TV executives experience the game using their mobile phones. To build that experience, I turned to Twilio.


Using Studio Flow in Twilio to create a dynamic SMS experience.
Using Twilio’s Studio Flow, I created a dedicated telephone # for The Algorithm which the Executive could “text”. The Executive (let’s call her “Alexa”) was the player in the “soundproof” booth.
Here’s how it went:
“Alexa, can you text this number ###-###-#### and say hi? The Algorithm is on the other end of the line and is ready to play.”

A sample of the Inventory questions given to the Executive in the Soundproof Booth.
After answering a few demographic questions called “The Inventory,” we were ready to play. At first, the questions started out easy:
“If you had to live in a $5 million pad, which neighborhood would you prefer?
A. Silverlake
B. Malibu
C. Downtown LA”
After Alexa locked in an answer, I asked her colleagues to make their prediction.
“Ok, [Alexa’s colleagues], you know her pretty well. What neighborhood would she choose?”
“Well, since Alexa needs to commute to Culver City every day, Malibu would definitely be the most convenient choice. And that’s all Alexa cares about … convenience. So we’re locking in MALIBU.”
“Alexa, what did you choose?”
“I chose Malibu!”
Yay. The co-workers got Alexa’s question correct. But so did The Algorithm. One of the “inventory” questions we ask the player at the beginning was “How many continents have you set foot in?” According to the data, women who have visited three or more continents are more likely to prefer waterfront property (i.e. Malibu).

Gradually, the questions begin to ramp up in difficulty, like this “Who do you lie to the most?” question above. And this is where The Algorithm really began to shine.

Using SMS interface to chat with The Algorithm
Humans start to second-guess themselves when things get fuzzy.
“I don’t think Alexa lies very often and definitely not to her colleagues. But if she did tell a lie, maybe it would be to her parents. Actually, wait … on second thought, I think it would be her friends.”
While the human players twisted themselves into knots, the Algorithm had a very simple framework to analyze this question. The data showed that for women living in Los Angeles, those who were 34 and younger chose “Parents” by a 3–1 margin. Women aged 35 and over, overwhelmingly chose “Friends” as the group they lie to the most.
It makes sense. If you’re in your twenties or early thirties, you’re probably not sharing every detail of your dating life and other exploits with your parents. But when you’re 42 and one of the Moms on your kid’s soccer team invites you to their cabin at Big Bear, you might just say you have plans.
In the end, The Algorithm won every time. Literally, every single time.
The executives at TV3 Ireland were so impressed they ordered the show straight-to-series.
And here’s the craziest part: over two seasons not one family beat The Algorithm. Not one.
This is just one example of how Artificial Intelligence can be used to bring an idea to life. Whether it’s creating an interactive experience for buyers or powering the underlying mechanics of a game, AI is a powerful tool that can help creators get their shows over the line.
