Computers Are Hard: AI Continues to Get It Wrong

The goal of artificial intelligence is to make our lives easier, but as a new technology, it still needs work. In some cases, as illustrated below, AI getting it wrong is good for a laugh.

The technology we all know and love is slowly taking over our world and our lives. Lay people, non-technical users, generally do not fully grasp what it takes to keep that technology working. And if you try to explain it to them, you’ll often be met with a blank stare. Why? Because computers are hard! This rendition of our increasingly comical series focuses on artificial intelligence. There are many pieces and parts to AI, and many forget that it’s still just a bunch of code written by an imperfect human. Recently, Amazon and a Scottish football (soccer) team gave us a reason to giggle at what can happen when AI simply gets it wrong.

Let’s start with the soccer team. Scottish football team Iverness Caledonian Thistle FC decided this year that, amid the pandemic and lack of fans in the stands, they would replace camera operators with artificial intelligence. Shouldn’t be too hard, right? Write a good algorithm, have a decent testing plan, hope that your training data sample is sufficient for your needs. You’re only training a program to follow a ball, how hard can it be? Show it a series of various soccer balls used by people of all different shapes and sizes and it will learn to follow that ball. 

Theoretically, it sounds like a good plan. In reality, Inverness fans ended up taking to Twitter to express their annoyance, irritation and sometimes a laugh. The AI-operated cameras kept mistaking the bald head of a line judge for the ball. If you watch the video, the camera struggles to distinguish between the ball and the referee’s head. It’s almost as if it’s a real-life expression of the Little Caesar’s “Crazy Calzony” commercial with the robot that can’t decide if it’s pizza or a calzone. The robot fries a circuit going back and forth, as it seems this camera does anytime the ball goes out of the frame or passes near the ref.

That one, while giggle-worthy, will not make you laugh as much as the Amazon story. Amazon is known to use AI to design and sell products on its site. But what happens when that AI creates a phone case sporting an elderly person wearing an adult diaper? Or toenail fungus remedies or Botox injections? There’s even a phone case with lines of cocaine and one depicting heroin. Those are just a few that are highlighted in an article on IFLScience.com

Medical terms somehow became increasingly connected to these phone cases, so if you click this link, you will see even more. Dental tools, an ultrasound probe, a girl with a hearing aid and more. The AI bot is supposed to generate phone cases based on popular image searches, but apparently Amazon didn’t restrict the images for use. They could have blocked medical terminology or self-care/self-diagnosis searches. They could have blocked the use of drug images. They could have set parameters around artistic searches, sports, nature, etc. But they didn’t. 

There are a couple of common areas of opportunity in both cases, but it is also important to remember that we cannot expect AI to do anything more than it’s programmed to do. It cannot account for things that it’s programming is not told to consider. Humans, therefore, must make as many considerations as possible. But we are an imperfect species, prone to making mistakes and cannot be tasked with understanding every single challenge along the way.

Computers are hard. Businesses are starting to learn that the development and use of certain technology cannot be rushed, but it’s going to take even more time for that to reflect in the products and tools we use every day. Artificial intelligence certainly has come a long way, but there is still a long way to go before its use doesn’t require human involvement. Which is why we suggest bringing in an expert for complex technology tasks, they can help ensure that you aren’t the next victim of an AI mishap.

About the Author

Pieter VanIperen, Managing Partner of PWV Consultants, leads a boutique group of industry leaders and influencers from the digital tech, security and design industries that acts as trusted technical partners for many Fortune 500 companies, high-visibility startups, universities, defense agencies, and NGOs. He is a 20-year software engineering veteran, who founded or co-founder several companies. He acts as a trusted advisor and mentor to numerous early stage startups, and has held the titles of software and software security executive, consultant and professor. His expert consulting and advisory work spans several industries in finance, media, medical tech, and defense contracting. Has also authored the highly influential precursor HAZL (jADE) programming language.

Contact us

Contact Us About Anything

Need Project Savers, Tech Debt Wranglers, Bleeding Edge Pushers?

Please drop us a note let us know how we can help. If you need help in a crunch make sure to mark your note as Urgent. If we can't help you solve your tech problem, we will help you find someone who can.

1350 Avenue of the Americas, New York City, NY