Meet Watson, your non-human resources manager

by TIM HORNYAK

YOU’RE ALMOST OUT OF college and it’s time to apply for a job. Hoping to score a full time position, you decide to try your luck with one of Japan’s most innovative companies. You’re competing against thousands of others, yet you trust in your winning personality. There’s just one thing: you’ll have to charm an artificial intelligence to land the job.

SoftBank has deployed IBM’s Watson AI platform to screen university applicants aiming to join the company in April 2018. As the first company in Japan to make such an announcement, the telecom giant was quick to say the system will reduce time that staff spend on the process by 75 percent.

This year, Watson is vetting 400 applications, which consist of typewritten answers to questionnaires. The platform is screening responses to one question, namely: “What are your strong points that match with SoftBank Values (No. 1, Speed, Challenge, Reverse Planning, Tenacity)? Please tell us about an episode that demonstrates your strong points.”

“First, we’re using Watson to evaluate job application forms more fairly by removing subjectivity,” says SoftBank spokesperson Rika Takahashi. “Second, to spend more time on communication with applicants.”

IBM CREATED WATSON (NAMED after its founder Thomas J. Watson) in an attempt to get computers to interact with humans in a natural way. Its ability to process natural language and draw upon encyclopedic knowledge earned it worldwide fame when it defeated human champions Brad Rutter and Ken Jennings on the U.S. quiz show Jeopardy! in 2011, winning a $1 million grand prize. Since then, the machine has shrunk from the size of a room to about three stacked pizza boxes, and it’s now accessible anywhere via the cloud. For SoftBank’s recruiting, Watson has been “trained” to discriminate between good and bad answers by analyzing 1,500 questionnaires pregraded by HR staff.

Watson is also one of the most persuasive examples of how what Big Blue calls “cognitive computing” is changing business. It has been deployed, with varying degrees of success, in everything from lung cancer screening to military procurement and even self driving buses. Japan’s Daiichi Life Insurance and Fukoku Mutual Life Insurance even introduced Watson based AI systems to sift and analyze documents related to payment assessments.

“‘Cognitive’ to IBM is really a new partnership between man and machine,” Jay Bellissimo, general manager of the Watson & Cloud Platform at IBM, told the New Economy Summit, an innovation conference held in Tokyo in April. “It’s really focused on accelerating, enhancing and scaling human expertise.”

“Some say SoftBank is a mobile phone company, but that’s wrong. We are an information revolution company.”

Able to read 800 million pages in one second, Watson can understand natural speech or text, assign contextual relationships to information and form hypotheses based on probability scores and algorithms focused on deep learning, a field of AI. As it ingests massive amounts of data, it can detect relevant patterns and assign them rankings according to their likelihood of fulfilling certain goals.

“If it gets something wrong, it will reweigh and the algorithms will do it again until it gets smarter and smarter and gets the right answer, ” said Bellissimo. “The real power is in the learning. Watson never forgets.”

In one example cited by Bellissimo, Watson helped Australian oil and gas firm Woodside by providing instant recommendations about oil rig design after going through 200 million bits of information based on 30 years of engineering experience built up at the company.

BUT AS AI APPLICATIONS proliferate, so do concerns that workers and their jobs will be squeezed out. Such prominent scientists and entrepreneurs as physicist Stephen Hawking and Tesla CEO Elon Musk have expressed fears that uncontrolled AI could pose a threat to humanity. On a more pedestrian level, Watson has struggled to live up to the massive hype surrounding it: After forming a partnership with IBM Watson in 2013 to battle cancer, the University of Texas MD Anderson Cancer Center put the project on hold late last year, its goals unmet despite an expenditure of over $62 million.

Despite a scathing recent report by Jefferies analyst James Kisner, who criticized Watson as too costly to develop compared to its earnings, the company has pushed on with the program as a core unit while it struggles to transition from traditional businesses such as servers and mainframes to cloud technology and cybersecurity.

For its part, SoftBank is happy to experiment with Watson and it has no doubts that AI is the future. The employee vetting platform is one of its latest AI projects following the launch of humanoid robot Pepper in 2015 and the acquisition of high profile robotics companies Boston Dynamics and Schaft earlier this year.

“Some say SoftBank is a mobile phone company, but that’s wrong,” SoftBank CEO Masayoshi Son told a shareholders’ meeting in June. “We are an information revolution company. A cellphone is just a device. From now on, we will be in an age where all infrastructure will be connected by information networks.”

Tim Hornyak is a freelance writer who has worked for IDG News, CNET News, Lonely Planet and other media. He is the author of Loving the Machine: The Art and Science of Japanese Robots.