Intelligent automation coming to IBM Cloud Pak for Data

Portfolio Lead, Watson Studio Data and AI, Cloud, IBM

What does a “journey” mean to you?  Perhaps a passage of sorts complete with adventures, overcoming challenges, and discovering new worlds?

At IBM, our long standing tradition of journey exploration has led humans to the moon and coined the term machine learning 50 years ago. Now we are helping organizations scale the ladder to AI to reap rewards in growth, productivity and efficiency with IBM Watson. This journey to AI mirrors the history of travel. In this article, I’ll explain how IBM Cloud Pak for Data accelerates the journey to AI and delve into  the ways AutoAI helps boost the speed of business returns.

Accelerating your AI journey with Cloud Pak for Data

Throughout history, human travel has continuously evolved to make our lives more convenient. In prehistoric times, humans got around only by foot until we discovered that we could cover more land on an animal. The invention of the wheel and the discovery that wood could float brought us carts and chariots, rafts, arks, and ships—enabling people to arrive at new destinations much safer and faster. Later came the industrial age and subsequent machines which increased automation in travel: trains, steamboats and ocean liners, automobiles, airplanes, and rockets.

Today, there is hardly a destination in the world that can’t be reached in 24 hours or less. Accelerating the pace of travel has helped us all reach greater heights. Likewise, on an AI journey, the desire and need for acceleration is just as critical.

Enter IBM Cloud Pak for Data.

A journey to AI involves a prescriptive approach of collecting, organizing, analyzing, and infusing data. With all the data available—such as images, text, device data, and relational data— to be  successful in prediction and optimization, businesses end up with a proliferation of tools and lack of data and analytic skills. In terms of speed, without a platform approach, this is akin to driving a car from New Orleans, Louisiana to San Francisco, California instead of flying. This cumbersome approach is not well suited to our fast-paced, experience-driven economy.

Cloud Pak for Data helps you speed through your AI journey on an integrated data and AI platform built on Red Hat® OpenShift® with IBM Watson® AI technology. With Cloud Pak for Data version 2.5, we are providing Watson OpenScale and AutoAI as part of the base and refreshing Watson Studio Premium, which consists of Decision Optimization, SPSS Modeler and Hadoop Execution Engine. The combination of open, modern data and AI platform with intelligent automation is a breakthrough concept—purposely designed to help a business extract higher value from AI; and build and scale AI with trust and transparency.

AutoAI, the supersonic air travel of AI, and OpenScale, the air traffic control tower

AutoAI is IBM’s AI for AI. By automating the development of AI, it’s the equivalent of air travel in the journey to AI, and that’s why it won an award for best innovation in intelligent automation. With AutoAI, you provide the dataset and indicate which column is the target—and the data preparation, feature engineering, algorithm selection, and hyper-parameter optimization is done automatically. AutoAI then finds the best possible machine learning model, guided by an AI system that provides the most promising flight paths and patterns at each stage. With Cloud Pak for Data, we’re on an airplane from the Big Easy to the Golden Gate Bridge. Together with AutoAI, we’re winging our way to San Francisco at supersonic speed. Imagine doing this by foot!

High volume air travel can be precarious on its own and for that reason, there are air control towers. An AI model that was created partly by automation can be risky in its own way. For example, is it biased, and can it be trusted? And how can we manage the volume of models? This is why IBM has an air traffic control tower for AI called Watson OpenScale. Watson OpenScale monitors model performance and accuracy, detects and helps mitigate the infusion of bias, and maintains the model's explainability.  Monitoring business outcomes and being able to explain them in human terms is crucial in ensuring fairness and productive use of AI.

Now that our trip has been reduced from 35 hours by car to under five hours by airplane, and made potentially even faster by supersonic plane, we’re able to enjoy San Francisco for a longer period and can also make additional trips more easily and much faster. This is how it can feel when you perform experimentation with AutoAI.

More iterations and better agility using ModelOps

Modern travel tools have made a journey from San Francisco to the French Riviera on short notice an exercise in agility. You can book your flight and hotel together. Mobile apps make it easier to check-in, clear customs, and head to the hotel. You’ll likely end up with more time to do the things and probably run into surprising delights—new discoveries in the narrow alleys, winning a game of roulette, and unwinding on a beach. It’s no wonder that, according to the Civil Aviation Organization, the number of world air passengers jumped from 2.628 billion in 2010 to 4.233 billion in 2018.

AutoAI and Cloud Pak for Data deliver the same kind of agility. Without them, much of model development involves tedious tasks that take time away from activities that deliver higher value, such as refinement, prediction, insights, and tuning results. These higher value activities require a collaborative, agile framework across teams from operations, data science, DevOps and application development. Powered by Watson Studio, Watson Machine Learning and Watson OpenScale, Cloud Pak for Data serves as an ideal foundation to implement ModelOps. It enables models to be pushed from a data science team to the DevOps team in a regular deployment and update cycle, aligned with continuous integration and continuous deployment to serve business.

Experiencing the new, open and diverse

Air travel, ridesharing, and mobile apps can offer today’s traveler an easy introduction to a new country and the freedom to cross borders in a short time. More of us are meeting people from different cultures, countries, regions, and hemispheres than we did just 20 years ago. We work together, share information and exchange ideas worldwide and all of us are benefitting from this diversity and openness.

Adding AutoAI to Cloud Pak for Data enables similar freedoms and experiences. The open architecture of Cloud Pak for Data ensures fast and easy deployment on the IBM Cloud, on other public clouds, on hybrid clouds—across any cloud. Because there is no lock-in, data scientists and the business are not only free to explore their AI discoveries, they can choose to collaborate with the community. Your AI models and insights benefit from the wisdom of crowds, production-tested quality, and the use of popular tools and approaches in a coordinated fashion.

Fifty years ago, we were just waking up to the possibilities of AI. However, IBM saw the potential early on and today, it is much greater than anyone could have dreamed of. IBM Research has always been on the forefront of AI and pushing its boundaries—even going so far as using AI to generate AI. Even more inspiring is that we get to partner with our clients around the globe and commercialize our ground-breaking research into AI that empowers and modernizes a business. To continue our conversation in AutoAI, optimization and ModelOps, join us in our 3-part Winning with AI Playbook Webinar. You  can also check out our popular Webinar playlist on upskill and deep dive data science. Note: AutoAI is available in Watson Studio Cloud today and is becoming available on IBM Cloud Pak for Data and Watson Studio Local later in Q4 2019.