FOR THE MASSES Strip District firm aims to make AI easy to obtain, easy to use
Artificial intelligence is the behind-the-curtains stuff guiding self-driving cars and other complex tasks. Developing the capability to do such things, Eric Xing says, is like building your own railroad.
Mr. Xing, co-founder and CEO of Strip District startup Petuum Inc., wants to change all that by putting sophisticated machinelearning capabilities at the fingertips of everyday users.
The company aims to create a product as ubiquitous as Microsoft Office word processing and spreadsheet software, while bringing the power of artificial intelligence to places such as the family doctor’s office, the shop floor of heavy manufacturers, and other sites not ordinarily associated with hefty computing muscle.
Petuum, which was founded in 2016, is getting ready to push its products out to the market and expects to need more people to help in that effort.
The 70-employee company recently raised $93 million in a second round of venture funding, a big infusion by any measure, which was led by a subsidiary of SoftBank Group Corp. The $93 million followed an initial round, which raised $15 million.
The new funding will be used to hire engineers and to build an internal sales and marketing team, said Mr. Xing, who expects the company’s workforce will reach 200 people by the end of 2018.
Petuum has three products in beta testing, and Mr. Xing said they will be easily adaptable to individual user needs. Other product details were not available, said Beth Klebacha, Petuum’s business manager.
Artificial intelligence or machine learning is the capability of a computer to change on its own through experience or data analysis. With surprising accuracy, music website Pandora and online retailer Amazon use artificial intelligence to predict what kind of songs we’d like to listen to and which things we’d like to buy, for example.
Like the word processing software programs used globally, Petuum’s goal is to make this capability affordable and easy to adapt — a “commodity that everybody can enjoy,” Mr. Xing said.
“Most industries do not have access to machine learning,” he said. “What’s needed is slightly boring, but something functionally more robust than artisanal artificial intelligence, which is the industry standard today.
“For daily use by a wide variety of industries, what’s needed is a VW, not a Rolls Royce.”
Mr. Xing, a native of Shanghai, China, is well-qualified for his role at Petuum. He received a bachelor’s degree in physics from Tsinghua University in Beijing, a doctoral degree in molecular biology and biochemistry from Rutgers University, and another doctoral degree in computer science from the University of California Berkeley.
Before Petuum was spun out of Carnegie Mellon University, Mr. Xing was a professor in CMU’s School of Computer Science.
His vision would expand artificial intelligence applications into areas such as helping a doctor find anomalies in a medical scan with more accuracy than any physician could do alone or crunching sensor data from jet engines to signal the need for routine maintenance or service.
“This is an exploding industry,” said Nick Nystrom, interim director of the Oakland-based Pittsburgh Supercomputing Center. “Genomic data, environmental data, wearable technologies — many, many modalities are coming together here.”
Computerizeddecision-support tools currently help guide doctors in making a diagnosis, but Mr. Xing sees a bigger role for his company’s product in medicine, which will be able to scan medical exams for the presence of a tumor, for example.
He is focused on offering products that are intuitive to use — “off the shelf, very, very friendly interface, which would make artificial intelligence a commodity instead of an art piece.”