In a small, nondescript office in the Flatiron District, some of the brightest minds in the world are teaching computers how to understand what they see. It is, perhaps, an unlikely place for the cutting edge of artificial intelligence to be built, but the technology they're creating will change the way you interact with the world.
Clarifai is building some of the world's most cutting-edge visual recognition technology. But, it's hard to tell the story of Clarifai without also telling the story of their founder and CEO, Matthew Zeiler.
After growing up in a small town in Canada, he went to the big city to study with some of the best minds in his field of computer neural networking. Neural networking is complex stuff, but at its core, it's building algorithms that try and copy how the brain works. Like a human, the more data a computer is exposed to, the smarter they get.
Zeiler first studied under Geoffrey Hinton, one of the pioneers of computer neural networking at the University of Toronto. Hinton would later go on to work at Google after they bought his company, DNNResearch, in 2013.
Zeiler went on to earn his PhD in machine learning and image recognition under the supervision of Rob Fergus at New York University.
When he was finished with school, Zeiler found his skills to be very much in demand, and he did what many of the world's best and brightest do: hopped on a plane to the Bay Area. There, he landed an internship at Google, where he worked with Jeff Dean at Google Brain, the company’s much-lauded deep learning project.
"I got to see to see firsthand how the big companies were doing AI," Zeiler said. “It was a great experience.”
Before he finished the internship, Google offered Zeiler the largest new graduate job offer in the company's history. Uncertain, he visited Microsoft, who doubled the offer — Google matched it. He met with Mark Zuckerberg, who also offered him a job.
“There was this great moment where I had this high reward and zero risk option," Zeiler said. "But, the tech we had back in New York was even more advanced than what we were doing in California.”
Zeiler then did the unimaginable. He quit his internship early and flew back East.
“I decided to follow my childhood dream and came back to New York to start Clarifai in 2013,” he said.
Clarifai makes a powerful image and video recognition API for developers and businesses. The company aims to make it easy for any company to integrate artificial intelligence and machine learning into their apps.
With just a little bit of input, Clarifai’s technology can teach a computer to recognize complex objects in seconds. With just a few clicks of the mouse, it’s possible to teach a computer to recognize the difference between, for example, a dog and a wolf. Or, the difference between a cup of coffee and a cup of water, or even the difference between a cup of Starbucks and a cup of swill from the corner deli.
This, of course, has significant ramifications for what we can expect our computers to be able to do in the very near future. It is, for example, possible to teach a baby monitor to know if a baby is awake or not. Or, after just a few swipes, you can train Tinder to recognize attributes you find attractive. The New York Public Library could digitize their entire catalog, and index it almost instantly.
The technology can, however, be used for less noble pursuits.
“We aim to provide the background AI to allow machines to understand every image and video to improve life,” Zeiler said. “We’ve had to turn down some opportunities that would not have met the criteria.”
Zeiler declined to offer examples or details, but it’s not hard to imagine such a powerful technology being used for all sorts of nefarious purposes.
For now, at least, the technology is being used for fairly benign purposes, and you may have already come across it. The photography community website 500px uses it to help index their photos, and companies like BuzzFeed, Unilever and Touchnote are using it as well.
Clarifai is still a very young company and secured a $10 million Series A in April of 2015. They currently have 33 employees in their New York offices, with another in the Bay Area.
“We aim to make it dead simple for everyone to utilize AI," Zeiler said. “We want to get this technology into the hands of the 7 billion people on the planet.”
Images via Clarifai