Dr. Bernard S. Meyerson, IBM Fellow, Vice President of Innovation, CHQ The Transformative Power of Technology The Roundtable on Education and Human Capital Requirements, Feb 2012
Dr. Bernard S. Meyerson, IBM Fellow, Vice President of Innovation, CHQ What skills are required in the new Technology driven World? The Roundtable on Education and Human Capital Requirements, Feb 2012
Service Growth: Pent-Up Change by Geography World s s Largest Labor Forces A = Agriculture, G = Goods, S = Service Nation Labor A G S % WW % % % 40yr Service Growth China 25.7 49 22 29 142% India 14.4 60 17 23 35% U.S. 5.1 1 23 76 23% Indonesia 3.5 45 16 39 34% Brazil 3.0 20 14 66 61% Russia 2.4 10 21 69 64% 2010 Shift to Service Jobs: US Case (A) Agriculture: Value from harvesting nature 2011 (G) Goods: Value from making products (S) Service: Value from IT augmented workers in smarter systems that create benefits for customers and sustainably improve quality of life. Japan 2.2 5 28 67 45% Nigeria 1.6 70 10 20 19% Bangladesh 2.1 63 11 26 37% Germany 1.4 3 33 64 42% CIA Handbook, International Labor Organization Beware Acceleration; What once required several years may now require only days.
A Need for Breadth & Depth: The New Division of Labor 15 Increasing usage of job descriptive terms 10 Expert Thinking (deep) 5 Complex Communication (broad) 0 Routine Manual -5 Non-routine Manua Routine Cognitive -10 1969 1974 1979 1984 1989 1994 1999 Based on U.S. Department of Labor Dictionary of Occupational Titles (DOT) Levy, F, & Murnane, R. J. (2004). The New Division of Labor: How Computers Are Creating the Next Job Market. Princeton University Press.
The Need for T-Shaped Professionals Broad across many Many team-oriented projects completed (resume: outcomes, accomplishments & awards) Many disciplines (understanding & communications) Many systems (understanding & communications) Deep in one system (analytic thinking & problem solving) Deep in one discipline (analytic thinking & problem solving) Deep in at least one
IBM: A Diversity of Disciplines Enables World Class Outcomes Behavioral Science Chemistry Computer Science Electrical Engineering Materials Science Mathematical Science Physics Service Science Science & Business & Engineering Management Technology Innovation Business Innovation Social Innovation Social & Cognitive Sciences Demand Innovation Economics & Markets
A History of Innovations that Matter Carbon Nanotubes (2003) High Temperature Superconductivity (1987) Deep Blue (1997) Cell (2005) Blue Gene/L (2004) Speech Recognition (1971)) Copper Interconnect Wiring (1997) Fortran (1957) RISC (1980) SiGe (1994) RAMAC (1956) Fractals (1967) First Petaflop Supercomputer (2008) One Device Memory Cell (1966) Nano MRI (2009) Relational Database (1970) Scanning Tunneling Microscope (1986)
Technology Provides Power for Rapid Change 1950s IBM RAMAC 350 Disk Storage Unit 2140 Lbs, 5MB max 2012 IBM Research demonstrated 12 Atom Storage
Continuous Innovation; Skills Beyond The AHA! AHA! Moment Never Undervalue Ongoing Innovation Disk Drives If IBM had not continued to innovate in this field, today s laptops would weigh approximately 250,000 Tons x2
Continuous Innovation; Skills Beyond The AHA! AHA! Moment Never Undervalue Ongoing Innovation Computer Performance If I grew in height from 1956-2011 at the same rate IBM grew computer performance, the moon would come around the earth and clip me in the ankles.
Computing Power: Exponential Growth Trends Source: http://news.cnet.com/ (Data from Top500.org)
Enabling a New Era; The Handling of Big Data The Roundtable on Education and Human Capital Requirements, Feb 2012
Defining BIG in BIG DATA In 2011 IDC estimated data created & replicated is 1.8 ZettaBytes* (1,800,000,000,000,000,000,000 Bytes) That is 1 Million times the total information held in all the Libraries of the United States Source: IDC Report June 2011 Extracting Value from Chaos
Dr. Bernard S. Meyerson, IBM Fellow, Vice President of Innovation, CHQ -Tackling REALLY Big Data- New Frontiers in Technology: WATSON The Roundtable on Education and Human Capital Requirements, Feb 2012
Jeopardy - The IBM Challenge 1997 - Chess A finite, mathematically well-defined search space Large but limited number of moves and states Everything explicit, unambiguous mathematical rules 2011 - Human Language Ambiguous, contextual and implicit Grounded only in human cognition Seemingly infinite number of ways to express the same meaning Watson is powered by: 10 racks of IBM POWER 750 servers 15 terabytes of RAM 2,880 processor cores operating at 80 teraflops Dr. Bernard S. Meyerson The Roundtable Financial Services on Education GTO 2011 and - DO Human NOT DISTRIBUTE Capital Requirements, Feb 2012 2011-04-05
What is Watson?
Through training Watson Evaluates and Selects documents worth analyzing for a given task. For Jeopardy! Watson has analyzed and stored the equivalent of about 1 million books (e.g., encyclopedias, dictionaries, news articles, reference texts, plays, etc) Too much irrelevant content requires unnecessary compute power
Learning Systems Roadmap to Meet the Challenge Autonomously & accurately identify essential features across multiple domains. Learning systems must understand context to disambiguate. Keyword Search Delivers lists based on keywords & human filters 1985 Watson Static Learning Systems Expert teams identify features across industries, create first commercial learning systems Today Dynamic Learning Systems Dynamic Data Corpus Expand Hypothesis Generation to different domains (leverage crowd-sourcing) Add Scorers for Different Input Modalities: images, video, voice, environmental, biological; leverage new devices & hardware acceleration Deeper Reasoning: Allow higher-levels of semantic abstraction. Leverage new hardware. Domain Adaptation Tools Autonomous Learning Systems Achieve understanding of natural language, images and other sensory information. Hypothesis and question generation across arbitrary domains; meta-heuristic to automate algorithm choices Biological Inspiration: Cognitive Process Understanding Future Greater Autonomy 18
Today s s vs. Future Systems: Calculating vs. Learning The Calculating Paradigm Archives Structured Data and Text Algorithms and Applications Static programming People Hypothesize, Determine what it means, Run other applications Input Same Output
Learning Systems Programming and Data transform to Learning and Intelligence Attributes Learn & Self-train (e.g. beyond programming) Interface Naturally & Accessibly with Humans and the World (including multimodal real time sensory input and output) Provide Insights, Create and Test Hypothesis Enable Better Outcomes A new class of systems that will be transformative
Dr. Bernard S. Meyerson, IBM Fellow, Vice President of Innovation, CHQ Watson in Healthcare The Roundtable on Education and Human Capital Requirements, Feb 2012
An explosion of clinical data is overwhelming healthcare providers How can we leverage it to address healthcare s common challenges? Facts per Decision 1000 100 10 5 Source: William Stead MD, IBM Global Business Services 1990 2000 2010 2020 Clinical Data Human Cognitive Capacity
Deep Q&A: Progress on Medical Queries 100% Winners Cloud Watson Y/E 2010 100% DeepQA Adapted for Medical Questions Precision Precision ~3 person-month effort Starting Point 2007 Eg. This is the most common cause of unilateral decreased vocal fremitus. 0% 0% 0% % Answered 100% 0% % Answered 100%
Healthcare Applications Evidence Profiles may be adapted to new domains Each dimension contributes to supporting or refuting hypotheses based on Strength of evidence and Importance of dimension for diagnosis (learned from training data) Evidence dimensions are combined to produce an overall confidences Overall Confidence Positive Evidence Negative Evidence
Watson a Roadmap Research / Demo Commercialization Future Technologies Won Jeopardy! Voice & Image Recognition Query & Dialogue 2007 2011 2011 2012 2012 2015
2011 IBM Technology Trends Survey Survey respondents identified Education as the industry with the biggest opportunity for IBM Watson s abilities Healthcare and aerospace/defense a close second and third, respectively.
Dr. Bernard S. Meyerson, IBM Fellow, Vice President of Innovation, CHQ How would you leverage a Genius at your side? The Roundtable on Education and Human Capital Requirements, Feb 2012
The Beginning of a New Era: Our New Co-workers? Dr. Bernard S. Meyerson, IBM Fellow, Vice President of Innovation, CHQ
Watson triggered Discussions just a few samples
Net-Net In the less than immortal words of Bachman-Turner Overdrive, You ain t seen nothing yet