Do Androids Dream of Quality?
Artificial intelligence improves data analysis, medicine and production robots
During its second match of the complicated board game "Go," Google DeepMind’s artificial intelligence (AI) program, AlphaGo, made a baffling move against one of the world’s best players, Lee Sedol.
"It’s not a human move," said commentator Fan Hui. "I’ve never seen a human play this move. So beautiful."1
Sedol stood up and left the game room. "He had to go wash his face or something—just to recover from it," said another commentator. Sedol returned, studied the AI’s move for 15 minutes before playing on and eventually losing four of five games.2
Many considered AlphaGo’s success a leap forward for AI cognizance, especially because it was speculated that AI programs were about a century away from mastering Go’s complexity, which requires near human-level intuition for victory.3
Sedol’s reaction to AlphaGo’s move might reflect what many people think about AI’s future role in their lives and businesses: What is it, and what’s coming next?
What is it?
AI is considered a subfield of computer science and has several definitions. On a basic level, it involves developing computers that can perform tasks associated with human intelligence, but designers’ goals for AI vary. Some work toward creating programs capable of simulating and explaining human reasoning—called "strong AI." Others’ aims are simply to design systems that behave similar to humans but don’t explain how we think—called "weak AI."
There are other distinctions. If a program is designed to generally reason, it’s called a "general AI." If it’s designed for a specific task—such as AlphaGo—it’s called "narrow AI."4, 5
Today, AI is used in areas such as healthcare and automated production processes. As organizations continue amassing vast amounts of data, machine learning—a branch of AI involving pattern recognition and learning without explicit programing6—is helping them spot trends faster and leverage more accurate predictive analytics.
"The people who figure out how to use those trends effectively, not foolishly, to increase their productivity, increase their capabilities and increase their quality are the ones who are going to win out in the long term," said Rodney Brooks, founder and chief technology officer of Rethink Robotics.7
Big data medicine
Market research firm International Data Corp. (IDC) estimates 50% of all apps will contain AI features by 2018.8 IDC also predicts 30% of healthcare systems will use machine learning with patient data and real-world evidence for personalized treatment regimens, including ones for half of physicians’ cancer patients, thus reducing costs and mortality rates by 10%.9
"Imagine there’s a new patient that has never been encountered by a particular physician before," said Eric Xing, a professor at Carnegie Mellon University in Pittsburgh. He asked whether a system could draw a connection between this particular patient and some previous patient treated by that system or elsewhere. "That could be made possible by an artificial intelligence system," Xing said.10
DermaCompare, for example, is a free phone app that uses AI and total body photography to help doctors and patients detect melanoma faster and more accurately. It does this by comparing a patient’s photo to his or her previous images or others in a database; creating a summary of data about the image, including any suspicious changes, and notifying the patient’s physician to review the information.11
Lior Wayne, CEO of Emerald Medical Management, explained that without AI, doctors must compare these photos with their own eyes. That’s why he views AI as a solution to some of healthcare’s most persistent challenges—time, money, accuracy and availability.12
Precision medicine is another area experts believe AI could revolutionize. Considerable variables, such as family history, lifestyle or genes, are involved in prescribing medicines. AI could make better sense of corresponding, and sometimes contradictory, relationships. This also includes drug development—understanding why a trial medicine affected one subset of participants differently than others.13
In manufacturing and production roles, robots can perform rapidly and maintain precision, but they’re limited by time-consuming programming requirements and a need for strictly controlled environments. AI and a method known as "deep reinforcement learning" could solve these issues.14
Fanuc, the world’s largest industrial robot manufacturer, is testing a robot that trains itself to do new tasks. One of its robots, for example, practiced taking objects out of a box and putting them into another. After eight hours of practice, it performed the task with a level of accuracy that’s almost the same had an expert programmed it.15
The learning process is accelerated if it shares information with other robots. If eight robots work together for an hour, they achieve the same learning as one that worked eight hours.16
Applying this learning style to robotics, however, is challenging. "Deep learning has made enormous progress in pattern recognition," said Ken Goldberg, a professor of robotics at the University of California, Berkeley. "The challenge with robotics is that you’re doing something beyond that. You need to be able to generate the appropriate actions for a huge range of inputs."17
Talk of automation and robots thinking like humans often makes people nervous. It conjures dystopian nightmares about being displaced by robots at work. In January, the World Economic Forum published a report that reinforced these anxieties, predicting robots will eliminate more than 5 million jobs by 2020.18
Yolanda Gil, a University of Southern California computer science professor, said people should rethink these fears. "AI systems in general are so limited," she explained. "One thing that people don’t understand is that they have no common sense and very little context about the world that we live in. They won’t replace or work in lieu of the real people. … It’s a way to add more capacity to the company without getting rid of the other jobs."19
—compiled by Tyler Gaskill, assistant editor
- Cade Metz, "The Sadness and Beauty of Watching Google’s AI Play Go," Wired, March 11, 2016, http://tinyurl.com/wired-go-ai.
- George Johnson, "To Beat Go Champion, Google’s Program Needed a Human Army," New York Times, April 4, 2016, http://tinyurl.com/nyt-ai-human-go.
- Kris Hammond, "What Is Artificial Intelligence," Computerworld.com, April 10, 2015, http://tinyurl.com/what-is-ai-cw.
- Nick Heath, "How Google’s AI Breakthroughs Are Putting Us on a Path to Narrow AI," Techrepublic.com, Jan. 28, 2016, http://tinyurl.com/narrow-ai-google.
- "Machine Learning," Whatis.com, February 2016, http://tinyurl.com/machine-learning-defined.
- Patricia Panchak, "Driving Artificial Intelligence and Robotics on the Factory Floor," IndustryWeek, April 13, 2016, http://tinyurl.com/iw-ai-robots.
- Gene Marcial "Zuckerberg Signals Wide Use of Artificial Intelligence on Internet for Medical Uses," Forbes, April 30, 2016, http://tinyurl.com/forbes-ai-medical.
- Tom Sullivan, "Artificial Intelligence, Cognitive Computing and Machine Learning Are Coming to Healthcare: Is it Time to Invest?" Healthcareitnews.com, April 22, 2016, http://tinyurl.com/ai-healthcare-learning.
- Jo Best, "AI, MD: How Artificial Intelligence Is Changing the Way Illness Is Diagnosed and Treated," ZDnet.com, April 27, 2016, http://tinyurl.com/ai-md-zdnet.
- Marcial, "Zuckerberg Signals Wide Use of Artificial Intelligence on Internet for Medical Uses," Forbes, see reference 8.
- Best, "AI, MD: How Artificial Intelligence Is Changing the Way Illness Is Diagnosed and Treated," see reference 10.
- Will Knight, "This Factory Robot Learns a New Job Overnight," MIT Technology Review, March 18, 2016, http://tinyurl.com/mit-robot-knight.
- Ker Sin Tze, "Dealing With AI and Job Displacement," Straits Times, May 2, 2016, http://tinyurl.com/ai-job-displacement.
- Stephen Desaulniers, "Chatbots Rise, and the Future May Be ‘Rewritten,’" CNBC.com, April 10, 2016, http://tinyurl.com/desaulniers-ai.
Who’s Who in Q
NAME: John B. Bowles.
RESIDENCE: Columbia, SC.
EDUCATION: Doctorate in computer science from Rutgers University in New Brunswick, NJ.
CURRENT JOB: Retired faculty member from the department of computer science and engineering at the University of South Carolina (USC) in Columbia. Bowles frequently taught the department’s introductory programming course and the senior capstone project course.
INTRODUCTION TO QUALITY: Bowles was project leader of the systems analysis group in advanced development at NCR Corp. The group was responsible for developing models and estimating the performance of computer systems while they were still in the design stage. Later, NCR decided to develop a fault-tolerant computer system, and his group got the job of analyzing the reliability of the system. For Bowles, this sparked an interest in reliability modeling. Years later in the 1980s, Bowles left NCR and joined the USC faculty. His first year at the university, he took a short course sponsored by the U.S. Air Force to teach faculty about reliability techniques and encourage teaching students how to design and build more reliable systems. This became his research focus and the focus of many courses he taught.
PREVIOUS JOB: After graduation, his first job was at Bell Laboratories—working on the U.S. Army’s Safeguard Program, an anti-ballistic missile defense system, perhaps the largest computer-driven system to be developed at that time. Many of the techniques for managing and ensuring the reliability and quality of large-scale systems were developed on that project.
ASQ ACTIVITIES: Served on the management committee of the Reliability and Maintainability Symposium, which has been sponsored by the ASQ Reliability Division for 17 years. Former treasurer of the ASQ Reliability Division and current treasurer for the division’s 2016 Accelerated Stress Testing and Reliability conference.
PUBLISHED WORKS: More than 90 journal articles, plus one book, several book chapters, conference papers and tutorials.
RECENT HONORS: A.O. Plait Award for Tutorial Excellence at the Reliability and Maintainability Symposium, and the Joseph M. Biedenbach Service Award from the USC College of Engineering and Computing. Member of the 2016 class of ASQ fellows.
PERSONAL: Wife, Ann; one step-daughter; and a new granddaughter.
QUALITY QUOTE: One of the most effective ways to improve the quality of any product or process is to determine what can go wrong and take the steps to keep those things from happening.
More Passengers Satisfied With Airline Services, Report Finds
Record profits driven by lower fuel costs seem to be allowing airlines to invest in the in-flight experience, resulting in higher passenger satisfaction, according to the American Customer Satisfaction Index’s (ACSI) 2016 Travel Report.
Airline satisfaction is up 4.3% to 72 on the ACSI’s 100-point scale, matching the all-time high set in 1994.
"Airlines have always been one of the lowest-scoring industries in the ACSI because the in-flight experience was miserable," said Claes Fornell, ACSI’s founder and chairman. "Historically, people were generally happy before they got on the plane. Now, that is changing. New planes, more options for in-flight entertainment, and the return of free snacks have resulted in higher passenger satisfaction."
Ticket prices have fallen, and ACSI data show passengers are perceiving a better value for their money. On-time arrivals also have increased, and cancellations have decreased. Despite the improvements, airlines remain among the lowest-scoring industries in the ACSI and at the bottom of the travel segment behind hotels, down 1.3% to 74, and Internet travel services, up 1.3% to 79.
For more information from the report, visit http://tinyurl.com/acsi-travel-report.
CASE CASE STUDY RELEASED A new case study detailing how a mining company used the define, measure, analyze, design and verify method to save millions of dollars has been released by ASQ’s Knowledge Center. Visit http://tinyurl.com/mining-dmadv to read it.
MERGER ANNOUNCED The ANSI-ASQ National Accreditation Board (ANAB) has signed an affiliation agreement with the American Society of Crime Laboratory Directors/Laboratory Accreditation Board (ASCLD/LAB), merging ASCLD/LAB into ANAB. Both organizations provide accreditation based on international standards for public and private sector crime labs. For more on the merger, visit http://anab.org/news.
Customer Satisfaction With Big Banks Continues to Grow
Big banks have significantly improved in overall customer satisfaction, while mid-sized banks have declined and regional banks have plateaued, according to the recent J.D. Power 2016 U.S. Retail Banking Satisfaction Study.
Customer satisfaction with big banks, in fact, has now risen for six consecutive years, seemingly driven by their improved digital offerings, more engaged personal interactions and stronger connections with growth segments of the population, the study concludes. Satisfaction with mid-sized banks dropped for the first time since 2010.
The trend of growing satisfaction with big banks "puts mid-sized banks most at risk," said Jim Miller, senior director of banking at J.D. Power. "Regulatory costs have made it difficult for them to invest in strategies to compete with larger rivals, and unless they take proactive steps to change course, we expect this to result in consolidation in the mid-sized bank marketplace."
For more details from the study, visit http://tinyurl.com/jd-power-bank-study.
Report Ranks Top Concerns for Supply Chain Executives
Dealing with limited facility capacity and resources that could strain supply chains, recruiting and retaining employees, and the growing complexity of products are the top three concerns pressuring supply chain professionals today.
That’s according to a new survey report by American Production and Inventory Control Society (APICS) and Michigan State University.
The report, "Supply Chain Issues: What’s Keeping Supply Chain Managers Awake at Night," details other challenges supply chain executives face, including compliance, cost and purchasing issues. The executives interviewed represented more than 50 organizations.
To access the report, visit http://tinyurl.com/apics-study.
GLOBAL STATE OF QUALITY RESEARCH
More Organizations See Quality as Strategic Asset
Fewer organizations view quality simply as a "check-the-box" activity and instead view it as a competitive advantage and strategic asset, according to the ASQ Global State of Quality 2 Research: Discoveries 2016 report.
Only 14% of respondents said quality is mainly a compliance activity within their organization, compared to 22% in the previous version of the research conducted in 2013. Of respondents to the 2016 survey, 36% said their organization views quality as a competitive differentiator and a strategic asset, up from 22% in 2013.
Given that more organizations view quality as a strategic asset and competitive differentiator, quality departments are now more visible within organizations, which has resulted in increased quality measurements and reporting, the report says.
The ASQ research also examines the state of quality and continuous improvement worldwide, providing organizations insights into gaps and opportunities. The latest research builds on the inaugural 2013 research, which provided the first-ever view of quality and continuous improvement on a global scale.
Discoveries 2016, released last month at ASQ’s World Conference on Quality and Improvement, provides an even deeper analysis and gauges progress since 2013. That report also addresses topics important to enhancing quality, including customer influence, trends in industry standards, training systems and governance structures. The report also identifies similarities and differences across nations and regions.
In addition to the quantitative and qualitative Discoveries 2016 report, ASQ and its research partner, the American Productivity and Quality Center, developed three Spotlight Reports that provide deeper analysis of topics covered in Discoveries 2016, including the role of big data in quality, the function of quality in innovation and the need for key performance indicators for supply chain success.
All Global State of Quality 2 Research reports can be accessed at no cost at globalstateofquality.org.