Dark patterns powered by machine learning: an intelligent combination

Like millions of others, Netflix r̶e̶c̶o̶m̶m̶e̶n̶d̶e̶d̶ autoplayed The Social Dilemma documentary to my iPhone, and it made an impression.

As someone who is involved in the industry of designing digital products and services I thought I knew all the ‘dark patterns’ that user experience (UX) designers use to manipulate users.

A dark pattern is a user interface that is intentionally misleading you.

Dark patterns in sales

A recent dark pattern I can’t escape is Youtube asking me to become a paying member. It takes over the screen asking if I want to subscribe for a fee and remove ads. But each time it asks, it asks me in a different way, with different ways to say yes or no, switches buttons around—anything it can to get me to say yes.

But this dark pattern is not designed by a user UX designer. It’s powered by artificial intelligence and machine learning.

The algorithm's goal is to optimize the ‘dark pattern’ to get you and millions of people like you to sign up—and here is the kicker, I believe it will win.

Here is how the logic works.

Goal: Get user (that’s me) to signup

• Regular power user — opens youtube once a day spends 20mins watching videos
• Consumes 8 ads per session but skips each ad before the ad finishes
• Closes Youtube during ad—bounces but returns within 10mins
• Request fullscreen takeover asking to signup and remove ads
• Request design A and of 18 possible designs dismissed, users who dismiss request A often respond to request C, next time run design C
• Increase ad play directly after dismissing the request to remove ads
• Continue until the user signs up

I want to reiterate this is not a person deciding this process, it’s a learning algorithm.