The CEO-in-waiting of analytics software firm SAS, Oliver Schabenberger, has set out the company’s strategy to help clients bring more automation to their data science and AI efforts.
“It’s ironic that the job that requires a lot of that automation is data science itself,” said Schabenberger, taking to the stage in Sydney in a double breasted blue blazer with bright red buttons.
SAS’s chief operating officer and chief technology officer is, according to a Bloomberg report, being groomed by the company’s long-time CEO and founder Jim Goodnight as his successor. This was not mentioned by Schabenberger, but local representatives did confirm the Bloomberg report.
Schabenberger joined SAS in 2002 after stints as a statistics professor at Michigan State University and Virginia Tech. His first role was developing statistical algorithms and models.
“For years we have worked on handcrafted models for object detection, facial recognition, natural language interaction and so on," he said. "And despite honing those algorithms, by the best of our species doing the best we could possibly do, the performance of those algorithms does not even come close to what we can accomplish today with a different approach."
Data scientists spend too much of their time manipulating, merging and cleaning data, to ready it for AI models and analytics Schabenberger said.
“They need AI in order to do their work in AI. To automate data preparation to make it more repeatable, use pattern recognition and machine learning to field the variable to obfuscate personal data and so on," he added. "We’re dealing with more varied data, unstructured data, and increasingly complex models. Automation can help find the best model, find the parameters, and automate the entire modelling pipeline."
By automating more aspects of the AI pipeline, SAS hopes to ‘democratise AI’. Building machine learning models will be within the grasp of not only data scientists but others in a business, including the C-Suite, Schabenberger said.
“Allowing everyone who touches data to consume and to produce analytics. All should be able to consume and produce right time insights,” he said.
Despite being one of few data analytics providers for many years, SAS now finds itself competing directly with the likes of Amazon Web Services, Google, IBM and Microsoft. Those players are all offering easy-to-implement AI tools, with which machine learning models can be quickly spun up.
To maintain its position, in March SAS announced it will be investing US$1 billion in AI over the next three years, mainly in software innovation, education, and expert services.
“The reason SAS tops the revenue list for advanced analytics for the last five years is that SAS solutions are built on a foundation of machine learning and deep knowledge of analytics. These are part of SAS’s DNA,” said Dave Schubmehl, research director for AI at IDC.
“Combining SAS’ knowledge and technology with its continued push to innovate in computer vision, NLP and deep learning will drive further adoption of AI across multiple industries. And it will help companies interested in AI – whether early in their AI and analytics life cycle or more mature.”
Schabenberger added that with SAS businesses could more easily scale up AI projects, and apply the technology across their businesses. He referred to AI as ‘Analytics Integrated’.
“We have to solve how to integrate it into business processes and operations. How to take AI from a science project to creating value for an organisation. From a science project to operational excellence,” he said.