We’re Now in the Era of DIY AI

December 19, 2018 - 7 minutes read

AI is transforming every industry. Want to know how machine learning will affect yours and how you can start using it to your advantage? Download our Free Machine Learning Whitepaper right now!

It wasn’t too long ago when artificial intelligence (AI) was a term you’d only encounter in works of science fiction. But with vast improvements in computing capabilities to understand speech and recognize images, this is no longer the case—AI is very much a viable reality.

And underneath the hype surrounding AI developments at tech titans like Amazon, Google, and Facebook, there is something special going on. In a bid to foster and recruit the best and brightest minds into their AI departments, these companies have released a plethora of tools that let amateur hackers and hobbyists utilize the same technologies these organizations are building the future with.

As a result, we’ve entered a new renaissance. Welcome to the era of DIY AI.

Identifying Plant Diseases on Sight

Sometimes, you’ve got to stop and smell the roses. That’s exactly what Shaza Mehdi, a high school senior from Lawrenceville, Georgia, was trying to do in her front yard. Except there was one big issue: her rosebushes were quite prone to sickness.

A big fan of Star Trek, Mehdi wondered why her phone couldn’t act like a tricorder (a handheld device from the series used to scan and analyze a surrounding environment) to diagnose her plants’ ailments. Soon after this thought, Mehdi and her friend Nile Ravenell began teaching themselves about coding neural networks in Python from online forums and YouTube videos.

Eventually, Mehdi found a tutorial showing her how to implement a Stanford researcher’s neural network which could identify skin cancers as well as board-certified dermatologists. She downloaded the software necessary to recreate this experiment. Initially, all it could do was recognize everyday objects like toilets and doors. But after Mehdi went through the painstaking process of collecting and feeding 10,000 labeled images of sick plants to the neural network, she knew she was onto something.

In 2017, after refining the neural network, Mehdi was ready to put plantMD, her resulting app, to the test. She focused her phone on an ailing grapevine outside of her house. After a few seconds, the words “grapevine anthracnose” popped onto the screen. Mehdi finally had her very own tricorder. Inspired by her work, she’s now a freshman in computer science at the University of Georgia.

AI for Art’s Sake

Arguments happen. And sometimes, the best way to prove your point is with action, not words. While Robbie Barrat was attending high school in rural West Virginia, he got into a debate with friends over whether computers could be creative. Sick of the back-and-forth, Barrat set out to teach himself code and create an AI capable of rapping.

After training his neural network on Kanye West lyrics, he achieved his objective. Here’s a sample: “I’ma need a fix, girl you was celebrating / Mayonnaise colored Benz I get my engine revving.” His friends were absolutely thrilled by the results. His teachers… not so much. “[They] got a little bit upset because the neural network was extremely profane,” explains Barrat.

It wasn’t long before Barrat’s talents were recognized by fellow techies. His AI rapper landed him an internship with a San Francisco developer of self-driving cars. Today, Barrat works at a biomedical lab at Stanford University, where he trains neural networks to identify molecules that have potential for medical treatments. But his heart still lies in making art with AI.

One of Barrat’s main hobbies these days is feeding AI photos and videos of fashion shows to produce new outfit concepts. He’s currently working with a designer to make some of these designs into reality.

AI-Powered Dry Cleaning

Dry cleaning is serious business in Japan. And for Daisuke Tahara, whose family owns eight dry cleaning facilities in a small town of roughly 50,000, finding great employees can be challenging. Looking for a solution to this conundrum, Tahara realized that computers offered some serious potential, particularly with logging and tracking orders.

But his employees struggled to adapt to the self-taught coder’s new system implementation. So Tahara figured, why not use machine learning to automatically check in a customer’s clothing? Using 40,000 photos of different articles of clothing, he set about training a neural network to accomplish this.

This past July, Tahara debuted his creation. Customers come into his shop and lay their garments on a table. From there, an overhead-mounted camera identifies all articles of clothing. With this great success in automation, Tahara has spared little time to celebrate; he has bigger plans: “I want to open a store with only the system and no staff.”

Innovation Can Happen Anywhere

With numerous sources and tools now available on the Internet, anyone can start innovating with AI today. “High school students can now do things that the best researchers in the world could not have done a few years ago,” says Andrew Ng. Ng is a famous AI researcher who has spearheaded initiatives at big tech companies like Google and Baidu.

He’s also well known for creating some of the first AI courses available to everyone via the Internet. It’s Ng’s hope that more amateur techies and AI enthusiasts embrace the knowledge now available online to create solutions to problems they face every day.

From the examples in this story, it sure looks like he’s getting his wish. What ideas do you have for AI? What do you think the technology could accomplish? And what’s stopping you from being the one to make it happen?

Tags: , , , , , , , , , , , , , , , , , , ,