What has a bigger impact and is more disruptive on the enterprise than the Internet, IoT, mobile/embedded smart devices, cloud, social networking, social media, 3D printing, augmented/virtual reality, robotics, and drones? Machine Learning (ML). Your enterprise will live, grow, and die from ML. ML provides huge competitive advantage for those strongly embracing it. What are you doing about it? How much do you know about ML?
What is the ML temperature globally?
I saw ML as a focus at the 2014 World CIO Forum where I was vice-chair and again at the 2015 World Computer Congress conferences where I keynoted. In the presentation by Korean researchers, they talked about a hybrid delphi ML prediction system to anticipate global trends years in advance and thus making the necessary investments. Korea already is top-ranked in e-government, education, innovation, hardware components, and percentage spent on R&D at nearly five per cent of GDP. They are also transforming their economy to be algorithm-based and away from hardware. ML is a serious topic at the United Nations in broader discussions in July and October, where I spoke and arranged a briefing with the world’s top authority Pedro Domingos to UN agency ICT heads. At the CIO CITY European Union CIO of the Year Awardees interactive panel session, Audi CIO Mattias Ulbrich, spotlighted machine learning as the next big wave!
What are ML-related use cases, forecasts and warnings?
IFIP (International Federation for Information Processing), founded by UNESCO and representing nearly one million scientists and professionals organized into associations reaching more than 90 countries, endorsed a global ban on autonomous weapons at their IFIP General Assembly. This added to an open letter supported by luminaries Stephen Hawking, Elon Musk, Steve Wozniak, Skype founder Jaan Tallinn, Noam Chomsky of MIT and 20,000 AI scientists and professionals announced at the 2015 International Joint Conference on Artificial Intelligence (IJCAI) in Buenos Aires, Argentina.
An array of notable luminaries signed an open letter from the Future of Life Institute calling for AI systems to do what we want and be beneficial.
Elon Musk spoke on ensuring we are not a biological boot-loader for digital super-intelligence. Bill Joy provided similar sentiments. Bill Gates in his REDDIT AMA spoke about his concerns with super intelligence matching others and opportunities in ML advances. “Even in the next 10 problems, like vision and speech, understanding and translation will be very good,” said Gates. “Mechanical robot tasks like picking fruit or moving a hospital patient will be solved. Once computers/robots get to a level of capability where seeing and moving is easy for them then they will be used very extensively. One project I am working on with Microsoft is the Personal Agent which will remember everything, and help you go back and find things and help you pick what things to pay attention to. The idea that you have to find applications and pick them and they each are trying to tell you what is new is just not the efficient model – the agent will help solve this. It will work across all your devices.”
MIT professors, Brynjolfsson and McAfee, believe we are in a second machine-age driven by smart-machines stemming from advances in AI, explosive growth in computing and communications, and the digitization of everything. The evidence is ample with driver-less cars, cell-reported traffic patterns, robots scanning and understanding environments, Microsoft HoloLens, Microsoft Skype language translation, computers writing email replies/reviews/resumes/grading essays, computers creating poetry, music, and classical art, computers deciding startup investments and even having a special decision board seat in an investment group. In the first machine age, increasing productivity resulted in growing jobs and income. In the second machine age, productivity is decoupled from jobs and income–a few employees can create products/services for virtually unlimited customers and at little cost. Good examples as of December 2015 are the nearly 150 unicorns which are startups worth more than one billion USD and in aggregate over .5 trillion USD. When you look under the covers of these unicorns, there is ML.
There are now a number of reports predicting the seriousness of the outcomes of ML. For example, McKinsey indicating that 45 per cent of jobs will be automated. Equally disturbing findings in a report on thematic investing and creative disruption from Bank of America and Merrill Lynch. Moreover, there are now “multiple” daily discoveries announced such as ML providing better diagnosis than expert doctors, beating humans in image-recognition accuracy, providing a more accurate analysis of galaxies, providing market/sector/stock trends, and Toyota announcing a billion USD research investment in Silicon Valley. Enlitic in their medical imaging ML system beat out a panel of top radiologists in detecting cancer and in another test spotting minute fractures not possible by humans alone. This list goes on and on essentially influencing every aspect of our lives.
Understanding the future impact, you will find ML-capabilities or support in new computing chips/technology: Nvidia Tesla M40, Nvidia Jetson TX1 development board providing mobile machine learning supercomputing, Qualcomm Zeroth technology, IBM TrueNorth neuromorphic chips with 1 million neurons and 256 million synapses (human brain has 100bn, 100trn), Micron Automata, and a diverse array of DARPA-funded projects (SyNAPSE, ElectRX, BRAIN, RAM). Interestingly the capabilities of your smartphone exceed the Cray supercomputers from the late 1990s’. This combined with GPUs and FPGAs making these ML developments possible. ML systems are demonstrating indicators of self-awareness, beating the average human for college entrance exams in Japan (announced in November) and soon gaining entry to the MIT/Harvard of Japan Tokyo University, displaying self-taught language skills of a 4-year old, learning from scratch how to master video games, allowing the augmented reality explorations by NASA, enabling adaptive security solutions at the application level, providing creative solution brainstorming of data rich opportunities and problems too complex for humans alone, providing from major vendors and startups open source access and tools (examples: SystemML, DMLT, TensorFlow, H2O.ai, Cafe, Theano, Torch, ConvNetJS, Deeplearning4J, Neon, Brainstorm) to advanced ML capabilities available even at the individual smartphone level. This is producing the Knowledge Society underlying the United Nations 17 Sustainable Development Goals and the ITU World Summit on the Information Society 11 action lines. In November, the US Commerce Department announced the Commerce Data Service to create data services on top of government big data allowing large and small businesses to capitalize on the ML revolution to grow their businesses without needing data scientists. Ultimately this will drive the US economy and jobs.
Pedro Domingos provided these mega-trends for my recent October keynote at the World Computer Congress:
- The transition from computers that are programmed by us to computers that learn on their own. This is enabled by big data, and in turn enables the personalization of everything, from medicine to shopping, and the increasing automation of every function in an organization.
- The automation of scientific discovery. Increasingly, each step of the scientific method, from gathering data to formulating hypotheses, is carried out by computers. This enables, for example, new drugs to be discovered at a much faster rate than before.
- The replacement of white-collar workers by machines, not just blue-collar ones. Routine intellectual work can increasingly be done by AI; what’s hard to replace is physical dexterity, common sense, and integrative intelligence.
- The transition from deterministic to probabilistic computing. From hardware to software, rigidly deterministic computations are giving way to probabilistic ones, enabling faster, cheaper, lower-power, larger-scale, more ubiquitous, more flexible, data-driven information systems.
- The rise of evidence-based X, where X includes medicine, policy-making, development aid, and ultimately all important societal decisions. Instead of guesswork and mixed results, we have randomized controlled trials that quickly weed out what doesn’t work from what does.
Last year I was invited to a special educational research workshop funded by the National Science Foundation. Grady Booch, a chief scientist at IBM, made these assertions:
- The mind is computable.
- The cosmos itself may be digital.
- Privacy is a recent illusion.
- Computing is a moral activity.
How you get educated and start preparing?
ML is easy to implement with the drag and drop tools readily available from all vendors such as in Microsoft Azure and seen widely entrenched in familiar rapid adoption environments such as Windows 10 with more than 110 million devices. Cortana, the Microsoft digital assistant, arrives in Canada providing executives insights into what is possible. HoloLens developer edition is being made available in early 2016.
Microsoft is open sourcing their big data, big model, distributed machine learning toolkit known as DMTK to developers on GitHub including easy APIs, parameter server framework, large-scale topic modeling, word embedding, and multi-sense word embedding. The DMTK can work with small computer clusters instead of the thousands required before.
Drilling deeper with Microsoft, you have ML:
– Bing producing predictions (World Cup, English soccer results, NFL games, top awards for the Oscars, etc.).
– Cortana analytics transforming business apps, custom apps, sensors and devices data into intelligent decisions for manufacturing, financial services, retail, healthcare, sales/marketing, customer/channel, operations/workforce.
– it is embedded in Outlook and Office products, business products and services, tools for enterprise administration and for developers, it is everywhere. For example two new features in PowerPoint are Morph and Designer, both designed to “easily” make presentations more compelling and reducing the required work. Designer uses ML.
– Azure ML providing diverse and accessible tools for creating prediction models and analytics.
– Project Oxford announcing the latest public beta for a tool that helps recognize emotions in pictures – so useful for marketers building on tools for speech, vision, language understanding, spell check, automatic video editing, speaker recognition, customized speech recognition for challenging environments, face detection tool, and more. In addition, you can have fun and try the demos.
– essential to IoT and Surface hardware such as Surface Pro 4 and Book
– Visual Studio working on integrating R
– Training systems to look at images the way people do
– Xiaoice virtual assistant with emotion taking China and Japan by storm and being converted to English. It remembers past interactions such as breakups with friends, identifies photos, carries on engaging real person like discussions, can act as a shopping buddy and moves into the realm of the movie Her
What is particularly compelling is there is emerging work on a Master Algorithm that takes the best of all five major tribes of machine learning creating an uber universal ML system. Leading this work is Pedro Domingos, who has achieved the data science Nobel Prize, Test of Time award, best paper awards and this year at the world’s largest AI conference, took the top prize for a system that is up to 10 “billion” times better than anything available today. His work underlies what you see from all the top companies such as Microsoft, IBM, Facebook, Baidu, Amazon, Apple, Google. He just published a book on the Master Algorithm that launched as a best-seller and is now doing a free streaming interactive talk with the non-profit ACM, the world’s largest association for research, education, innovation and conferences like Applicative and Queue mobile publications for IT pros and developers. The streaming session will provide background on the five top schools of machine learning and then talk about the Master Algorithm—an excellent tutorial for IT Pros and developers wanting a good fundamental understanding of ML and where it is heading. To participate go to this page to register for the November 24th “The Five Tribes of Machine Learning (And What You Can Learn from Each),” presented at 12 pm ET (11 am CT/10 am MT/9 am PT/5 pm GMT). If you miss the event, you can get it on-demand in the ACM Learning Center.
What is the reality?
Going forward, we live in a world where there are:
- Unlimited computational resources and connections.
- Pervasive computational thinking.
- Whatever the future, it will depend on computing.
- Everything is recorded, nothing is forgotten.
- Organizational, geographical boundaries disappearing.
- Moving towards a master algorithm — universal learner.
- IoT -> global digital mesh -> planetary nervous system -> ML -> Knowledge Synthesis of Everything
What are the big questions?
There is an impending digital quake where by 2030 more than 80 per cent of companies and jobs change.
What are the economic implications?
What is the social impact?
What will the world look like?
What are the intended and unintended consequences?
Is there a need for ICT accountability, ethical conduct, credentialing which equals professionalism?