AI Makes A Play Into Pranks

Here at PYMNTS over the last few years, we’ve had no shortage of things to write about artificial intelligence (AI) and machine learning — and the many wondrous contributions these technologies can make to ensure that our payments and commerce experiences are safer, smoother and more customized.

In much the same way that eight-year-old Amazon Alexa is no longer the “shiny new toy” in the space, the near-ubiquity of AI has resulted in a “What have you done for me lately?” frame of mind.

This week, for example, there is a surprisingly good answer to that question, as AI is being used to improve the creative and innovative endeavors of pranksters. It’s a reality demonstrated this week when one bored inventor decided to use AI to exact revenge on his neighbors after municipal authorities dug up his sidewalk for “essentially no reason.” Why punish his neighbors for the city’s project? Because the lack of sidewalk has prompted people to walk across his lawn, making it impossible for him to replant it.

It’s a situation that this particular homeowner was not going to take lying down. Ryder (the given first name of the digital prankster) was already working on an AI-powered Raspberry Pi camera, training it to recognize dogs and eventually other objects, working in conjunction with a Raspberry Pi computer. But given that his front yard had become a pass-through for pedestrians, he decided to customize that technology to a higher purpose: using AI tech to recognize people walking on his lawn and blasting them with a spray of water from the sprinkler as punishment for their impertinence.

“There is no teacher quite like fear. So I’m using artificial intelligence to turn my sprinkler on only when people walk on my lawn,” said Ryder. “I made this for entertainment only. I wouldn’t recommend that you actually build anything like this yourself — I just wanted to make a video that I thought was entertaining.”

You can watch the video below. Spoiler alert: It is very entertaining. And by all accounts, the video shows that the prank got the job done — anyone who was hit with a blast of cold water quickly jumped off Ryder’s lawn, usually while shrieking.

And while Ryder’s prank is perhaps more inventive than most, AI tools have a long and glorious history in the hands of practical jokers. Telecom industry drop-out Roger Anderson, after a particularly aggressive run-in with an AI marketer, decided to strike back with AI tools of his own, designed to waste their time as aggressively as he felt they were wasting his.

With a few custom voice recordings, ​​he programmed an artificially intelligent robot to converse with telemarketers for as long as it could by creating an algorithm that says, “Hello? hello?” to trick the agent into thinking they have reached their intended human target. The robot then says “yes, uh-huh, right” to answer the telemarketer. Once the telemarketer gets suspicious, the bot is programmed to give increasingly inane responses to questions until the caller gets fed up and hangs up.

And if that all sounds like delightful good fun, but you don’t have the programming ability to set it up for yourself, we’ve got good news.

AI’s voice generation and modification functions, according to the esteemed experts in this area, are apparently the greatest technological advance in the institution of the prank phone call since push-button phones replaced rotaries.

“With this software, it wouldn’t be long before pranking becomes as easy as pressing a button. I am sure there are now programmers whose main project is to develop a fully functioning AI that allows you to prank people with just one click,” the team at Tech Wonders wrote.

A Word of Warning

AI is a great assistant for those looking to prank their friends and family, but it is important to note that it’s only truly useful in carrying out pranks that a human designed — it is not yet at the level where it can generate pranks on its own.

Not because it hasn’t been tried — it’s just that AI hasn’t come up with particularly good, or even sensible, ideas in this arena. Those who tried to use a neural net GPT-2 to generate some April Fool’s pranks, for example, got some … odd suggestions.

On the list: Glue a toothpick to the tops of grapes, put your fear of insects into a lemon, paint the refrigerator with red spiders, and, our personal favorite, spread the jelly eggs on the toothpicks to attract the jelly eggs.

It seems the neural network needs a bit more training, as it is perhaps not perfectly clear on the concept of what a prank is — or that it is something you do to someone else, not to yourself or a jelly egg.

But then, as countless innovators have told PYMNTS in the past about AI, this tech is only as smart as the person who programs it — and it can’t come up with original ideas on its own. That means generating pranks — which ideally require some degree of a sense of humor — may always be a bridge too far.