SAS Institute CEO and co-founder Dr Jim Goodnight has played a central role in the rise of data analytics in his 41 years at the helm of the software giant, but the billionaire businessman with the statistics PhD doesn't buy the AI hype.
"Everybody's going to claim that they do AI because that's the latest thing," Goodnight told Computerworld UK in Amsterdam at Analytics Experience, his company's annual European conference.
"It used to be cloud computing and it used to be big data and everybody said they're big data, and everybody said they're cloud, and now everybody has to say they're AI. Same for us."
Goodnight believes companies are taking advantage of the growing interest in AI to promote their brands, including IBM Watson, which he says is essentially a search engine whose intelligence "is mostly marketing".
"Analysts and reporters and editors are tired of talking about cloud," the six-foot-four Goodnight proffers in his soft Carolinian drawl.
"They have to talk about something new. It's happened my entire lifetime. Every four years there's a new buzzword that [everybody] is stuck with, and now it's AI."
AI may now be exploding into the public consciousness, but Goodnight says that SAS has been deploying it for a long time: "We've really been involved in it for at least 15 years. All of our models in fraud detection for credit card fraud are really AI models. They're built with neural networks. Anytime you allow a machine to make a decision a human could have made then that's called artificial intelligence, a very broad topic.
"It could be somebody writes three if statements that makes a decision that a human normally would and you can call that artificial intelligence."
His cynicism extends to the fanfare for machine learning, another data science buzzword.
"The whole word machine learning is all hype as far as I'm concerned," he says.
"You have a definitive model that you've defined, and you are trying to read through the data and improve the model by minimising the objective function. You do it over and over again until you can shrink that objective function, and then you have a model that you then use in production.
"But there is always a model behind this stuff. You don't just say 'Hey! Go learn'. So machine learning is really a misnomer. It sounds good. But don't get carried away about it. It is just very complex models that are being fitted."
Disruptive force of analytics
SAS famously rejected the temptations of an IPO or acquisition offers to become America's largest private software company in the business intelligence market, with revenue of $3.2 billion in 2016, following annual growth every year since SAS was founded.
Competition has increased since then, both from tech titans such as Google and more nimble startups. Goodnight believes that SAS stays ahead by focusing on its customers.
"We try to make sure we develop stuff our customers want," he says. "We not only listen to our customers, but we also try to understand where the market is moving so we can get there first."
In the opening session of the conference, Goodnight said that SAS analytics software is designed to help these customers "be the disruptors and not the disrupted."
He applies the same drive for disruption to the SAS business strategy: "That's what we try to be too. We've been around for 41 years because we are continuously improving our software and making changes and adapting to new problems that arise out in the field."
These improvements can emerge from schemes like company innovation labs where staff present new ideas and if the concept has business potential SAS will fund the development.
Among the proposals that went into production is Viya, a cloud-enabled in-memory analytics engine that was launched in 2016, eight years after the idea emerged. The architecture enables massively parallel computing, a type of computation that executes numerous calculations or processes simultaneously.
SAS defines parallel computing as a practice "in which certain computations are partitioned into independent smaller subcomputations. Each subcomputation is then processed on separate cores or processors simultaneously."
"That has been a huge challenge, because that's not how we learn these things," says Goodnight. "We learn basically a sequential process. You read the data in, you fill the compute this matrix, you invert that matrix and then you print out the results.
"How do you do that in parallel? It's not in any of the books. You have to figure it out, and that's what we've been doing for the last eight years, figuring out how to put all of our computations into parallel.
"There's some jobs that used to take 18 hours that we can run in four minutes now, so it's really helpful to those customers that do have large amounts of data."
Recent Viya adopters include sports analytics company SciSports, which applies analytics to event-stream processing in football matches to quickly identify anything that happens on the pitch.
"We're getting pictures that are 32 frames a second from TV cameras and that has to be analysed, and if you get 20 cameras around the pitch there then that's a lot of data to process every second," says Goodnight.
Event streaming has become a big part of the SAS business, particularly as the Internet of Things (IoT) is making data generation, collection and analysis possible at its point of origin in millions of edge devices, from household appliances to locomotives.
"We have moved a lot of our analytics into the event stream," says Goodnight.
"We take streaming data for about 400,000 events a second and we can process that and then we take that and stream it right into Viya so that you have your visual results from all that data movement.
"We're working with companies on building all the air conditioning controls, the heating controls, the electrical controls, where we monitor those 24 hours a day and then give the customer suggestions on how they could improve efficiency just by turning this nob or that nob or changing the damper that they could actually reduce their electrics cost. And also we use the data for preventive maintenance to forecast when something is going to get bad."
The data science skills gap
Finding the necessary data science skills to drive these efficiencies remains a struggle for most enterprises.
In a recent SAS (Statistical Analysis Software) survey of executives from 100 European organisations that was released in August, only 20 percent of respondents felt their data science teams were ready for the challenges of AI, while 19 percent said they had no data science teams at all.
It's a subject close to Goodnight's heart. His philanthropic focus on the analytics skills gap is a key reason why the 74-year-old chooses to continue his work in data science rather than put his feet up and enjoy the $9.4 billion dollar net worth that makes him the richest resident of his home state of North Carolina.
The education initiatives that he's driven include SAS Curriculum Pathways, a web-based learning environment used by more than two million teachers and students, and the Goodnight Scholars Program, a scholarship for STEM students from low- and middle-income families.
The scheme supports more than 200 undergraduates every year at his alma mater North Carolina State University, where the SAS story began in 1976, when Goodnight and his fellow PhD student John Sall conducted a research project developing a statistical analysis system for agricultural data that they developed into the company of today.
The skills gap that these schemes help bridge isn't felt as keenly at SAS than it is at most other companies, which Goodnight attributes to the business' compassionate working environment.
SAS' combination of fluid organisational structure, flexible working and perks including an on-site gym, day care facilities and free work-life counseling has earned the company a spot on Fortune's 100 Best Companies to Work For every year since the list was introduced in 1997.
The compassionate corporate culture that Goodnight fostered is said to be the inspiration for Google's famously friendly workplace.
"We don't have too much trouble finding talent, because we're one of the great places to work around the world," he says. "That helps us find talent. It helps us retain talent as well."
An optimistic take on GDPR and Brexit
Goodnight believes that concerns about how companies use personal data are somewhat overstated.
"You have GDPR here so if you don't like it just opt-out," he suggests. "By and large I don't see a whole lot of damage that can be done. They can pester you to death dropping cookies and every time you go to a website they hit you with the same ad you've seen a dozen times. When that happens just start clicking through and costing them lots of money."
To help firms prepare for the stricter requirements for data processing that GDPR will enforce, SAS has launched a 'one-stop shop' software to help firms locate and isolate identifiable customer data to ensure that it's protected.
Goodnight is relaxed about the impact of the regulation, but admits his company didn't support its implementation.
"Quite frankly our people in Brussels opposed it because we thought it would be a lot of effort to try and guarantee somebody's privacy and the right to be forgotten, so our representatives in Brussels were pushing against it," he says. "But we lost. But now we have the software to help companies uncover any use of data like that."
His scepticism about the EU makes him sympathetic to Brexit. Europe is the source of more than a third of SAS' global revenue and the UK is a key market, but Goodnight is unconcerned about the impact of Brexit, and admits he would have been a 'leave' voter.
"It doesn't bother me," he says. "I quite understand it, having Brussels tell you everything to do. I was with the English on that one."