Follow us on Twitter: @GoAfricaNetwork
Amelia, experts say, is “the next level up” from IBM’s Watson supercomputer.
In 1998, an independently wealthy New York University doctoral student gave up his assistant professorship in math and founded a company to build software that could mimic the human brain—a “cognitive” platform that could learn a job, not just search for programmed answers.
On Sept. 29, IPsoft—now a global company little known outside the world of IT infrastructure services—unveiled Amelia, which consulting firm Gartner calls “the next level up” from IBM’s Watson supercomputer and “a breakthrough innovation” that signals “the end of outsourcing as we know it.”
BY MATTHEW FLAMM SEPTEMBER 29, 2014 1:00 A.M.
The brainchild of IPsoft founder Chetan Dube, Amelia can digest an oil-well centrifugal-pump manual in 31 seconds—and give instructions for repairs—and do the job of a call-center operator, a mortgage or insurance agent, even a medical assistant, with virtually no human help. Fluent in 21 languages, she understands implied, not just stated, meanings, and improves her performance by hearing humans deal with questions she can’t yet answer.
Monday’s unveiling begins a push by IPsoft to show off Amelia to the business world. She is already involved in pilot efforts in five industries, including institutional banking, insurance and oil and gas.
Experts—and IPsoft—say that Amelia and cognitive technology in general mark a turning point in the automation of business processes and the start of a new wave of disruption.
“We are at the edge of a tectonic shift,” said Mr. Dube, the company’s chief executive. That shift will have its most profound effect on the global “knowledge worker” sector and the $9 trillion that McKinsey & Co. estimates it totals in employment costs.
“Our assessment, and our partners’ assessment, is that a third of that market segment is immediately addressable by technology such as Amelia,” Mr. Dube said. “That is a $3 trillion market impact at maturity.”
Raised in New Delhi, Paris and London, Mr. Dube built IPsoft into a profitable IT services business that manages infrastructure through automated systems, with offices in 10 countries and 2,000 employees. About 500 work out of the company’s downtown headquarters overlooking New York Harbor, and more than half of them are doing research and development supported by two Dube family trusts.
Mr. Dube declined to describe the source of his family’s wealth. He also declined to provide a revenue figure for IPsoft, which is privately held. In a 2011 article, India’s Economic Times reported it as $700 million.
IPsoft spent the first 10 years of its work on Amelia studying the human brain, and without writing a line of code. The approach centered on “neural emulation” and how the mind maps connections between objects, which Mr. Dube sees as fundamentally different from IBM’s focus on emulating the brain’s architecture with neurosynaptic chips.
Room to grow
He likens IPsoft’s efforts to people learning to fly by understanding the principles of flight rather than by imitating a bird (hence the product’s homage to Amelia Earhart). The company also insists that Amelia comes closer to human understanding than the biggest name in smart machines, IBM’s superstar computer system, Watson.
“Watson is very impressive, but its strength is big-data analytics,” said Ergun Ekici, vice president of emerging technologies at IPsoft. “It is not comprehension of what a person means when they talk to you.”
An IBM spokesman did not respond to requests for comment. But one expert advised caution when comparing the two systems.
“[IPsoft is] doing some interesting things, which are different from how Watson does things,” said Kathleen McKeown, director of the Institute for Data Sciences & Engineering at Columbia University, who is familiar with both Watson and Amelia. “It is apples and oranges.”
In addition, though Amelia is making her debut, she is not fully grown. IPsoft continues to work on her reaction to emotions, and there are limits to what she can infer from ambiguous statements.
“Contextual ambiguity—when the context ends and begins—that will be one of the challenging things,” said Jinho Choi, an assistant professor at Emory University who worked at IPsoft and continues to consult on Amelia.
Mr. Ekici said the company has been improving Amelia’s ability to handle ambiguity—and she asks questions when something isn’t clear. He also expects her to learn a lot more once she’s put to work.
In the meantime, analysts and business leaders are excited about Amelia’s potential. “It’s one of the first things we’ve seen that’s commercially available that actually learns,” said Frances Karamouzis, a Gartner analyst who praised the platform in a report earlier this year.
The cloud that hangs over Amelia, and over much of the technology industry, is what happens to the people who lose their jobs. News articles in India in advance of the platform’s launch have been largely about the threat it poses to the country’s outsourcing industry.
Last year, Gartner forecast the entire managed-services industry would see a 60% drop in costs by 2017, all from lower head count, owing to new technology.
Jobs killer or creator?
Some IPsoft clients, like Accenture, agree that Amelia will be disruptive, but say it can lead to the creation of higher-value jobs.
“In addition to efficiencies, it can enable you to do new things, like expand your services,” said Paul Daugherty, Accenture’s chief technology officer, giving the example of a manufacturer selling a piece of equipment. “Companies want to sell a service, not a product. Using cognitive technology, they can layer on services.”
Mr. Dube argues that Amelia will free professionals from mundane chores and allow them to focus on innovation. He also thinks robots and smart machines will soon be central to daily life.
“I’m convinced that in the next decade we will pass someone in a corridor and we won’t be able to tell if this is a human or an android,” he said. “The difference between science fiction and science is time.”
A version of this article appears in the September 29, 2014, print issue of Crain’s New York Business.