The Future of Data Science: Augmentation, not Automation


A decade ago the iPhone was a new device and there wasn't an app for much of anything. Competing on analytics was a new concept. And we were considerably less worried about artificial intelligence taking our jobs and taking over our world.

Boy, things have changed in the last 10 years. Those changes were the major theme of the speed keynote talks on the mainstage at the SAS Analytics Experience conference in Washington, DC, yesterday, featuring former US Chief Data Scientist DJ Patil; SAS CTO Oliver Schabenberger; and author, professor and consultant Tom Davenport.

Patil

"We live in a remarkable time," said Patil, leading off the lineup of speakers.

For instance, he said, we expect instant and customized news, real-time traffic information and mapping, car service on demand, one-day shipping, and many other instant conveniences.

Patil said that the US created the office of the US Chief Data Scientist with a particular mission in mind -- to responsibly unleash the power of data to benefit all Americans.

"Two words that were carefully chosen were 'responsibly' and 'all' Americans," Patil said. "We should think of technology as neither radical nor revolutionary unless it benefits every single person."

Patil listed several efforts underway to unleash the power of data in the nation. One effort in Miami Dade County, Florida was designed to improve data sharing among agencies to enable better crisis intervention by police officers. When the police force was provided with a spreadsheet of those suffering mental health issues, the police were better equipped to deal with crisis situations and to route mental health patients to safe situations instead of prison.

(right to left) Host Mark Jeffries poses questions to industry leaders Patil, Davenport, and Schabenberger.

(right to left) Host Mark Jeffries poses questions to industry leaders Patil, Davenport, and Schabenberger.

Other examples included grassroots efforts in precision medicine for cancer and genetic disorders, and device tracking data to help determine where cities should spend money to improve lighting and safety.

Patil said that people are the major force that enables data to fix problems.

Schabenberger

Schabenberger provided an overview of some of the emerging technologies around data science, analytics and artificial intelligence and reassured the audience that today's artificial intelligence does not work like the human brain, and tomorrow's AI won't work like the human brain either. But AI and related technologies do offer hope for greater automation in the future and that's something that we need.

"We are generating huge volumes of data," he said. "We are streaming and handling much more data than we need to. Devices are not often within network range and they have expiration rates." That's why one of the emerging tactics is to collocate data and compute. That is how you can create a wind turbine that can best adjust its own blades with data from the field, for instance.

"Our interest in machine learning reflects our need to process faster," Schabenberger said. "It's not a model driven approach. It's a data driven approach. So that appeals to us because it affords opportunities for automation."

Now, with technology like deep neural networks combined with big data, we are achieving accuracy in natural language processing, computer vision and more. So for instance, a machine can be created that can beat a human in a game of Alpha Go. And those same lessons can be applied to other situations, too. For instance, deep neural networks may be applied to help maximize the lifetime value of the relationship with a customer.

Davenport

Davenport's appearance coincided with the 10-year anniversary of his book, Competing on Analytics, and the publication of the new and updated edition of the book. It's been a decade of change for analytics, and so there was plenty of opportunity to update the book.

In his presentation, Davenport listed 10 analytical technologies that came of age within the last decade, including Hadoop, Spark, Python, text analytics, streaming analytics, and social media analytics.

"This was the era in which everybody figured out what Money Ball was," he said. It used to be that a couple sports teams employed an analytics person. Now every team has an analytics person, Davenport said. Now many more industries and organizations are deploying and competition on analytics.

Davenport provided insights into the different eras of analytics -- artisanal analytics (1.0), big data analytics (2.0); data economy analytics (3.0); and cognitive analytics (4.0) -- and some of the processes required for progress in each one.

Ultimately, all of the panelists agreed that the goal is to augment human work and decision making, not necessarily to automate it all.

"There is a great difference between automation and autonomy," Schabenberger said.

Jessica Davis, Senior Editor, Enterprise Apps, Informationweek

Jessica Davis has spent a career covering the intersection of business and technology at titles including IDG's Infoworld, Ziff Davis Enterprise's eWeek and Channel Insider, and Penton Technology's MSPmentor. She's passionate about the practical use of business intelligence, predictive analytics, and big data for smarter business and a better world. In her spare time she enjoys playing Minecraft and other video games with her sons. She's also a student and performer of improvisational comedy. Follow her on Twitter: @jessicadavis.

A2 Academy Fall 2017: 5 Hot Technologies You Need to Know

Are your career and your enterprise organization ready for tomorrow's analytics technologies? To help get you up to speed, we are offering a series of five webinars on Natural Language Processing, Machine Learning, Edge Analytics, Real Time Analytics, and Augmented Analytics.

QVC's Vision for Online Retail Goes Real-Time

Before Amazon or Google even existed, QVC had a vision of a world where you could buy a raincoat and have it delivered within hours. Here's a look at QVC's next vision and innovation.


Re: Perspective
  • 9/26/2017 1:04:36 PM
NO RATINGS

@Ariella: Maybe. But there are enough reasons for and against AI and automation in industry without having to resort to the "They took our jerbs!" argument. If a person's job is so replaceable and disruptable to begin with, then the industry has fundamental problems of fragility that are greater than any technology issues -- and those jobs have always been at risk for disruption, but we're only just paying attention to that risk now while setting up automation as the strawman. (Case in point: "Legacy" taxis vs. apps like Uber and Lyft)

Re: Perspective
  • 9/25/2017 9:40:36 PM
NO RATINGS

@Joe True, but I get the impression that even with some winners in the game, there will likely be more losers. There are some visualizations of the areas expected to take a hit from automation here: /www.bloomberg.com/graphics/2017-jobs-automation-risk/

Re: Perspective
  • 9/25/2017 8:35:59 PM
NO RATINGS

@Ariella: Some people will lose jobs. Some people will gain jobs. Some people will be able to transfer their skillsets and get upskilled/reskilled.

My take is that it's a red-herring issue. If the number of jobs is our chief and sole consideration, we can also get rid of tractors and electric light bulbs and the like. Inventions, if they're any good, always change up the job landscape. Disruption is a natural part of the cycle.

Re: Perspective
  • 9/25/2017 9:56:20 AM
NO RATINGS

@Joe What's your take? Do you think the academics are correct in their outlook?

Re: Perspective
  • 9/24/2017 9:00:05 PM
NO RATINGS

@James: That was definitely the theme of MIT CIO this year, as well as numerous other conferences/events I've attended. Only the academians have been the ones really concerned about job loss. The people in the private sector who actually do business for a living have a less dismal outlook, in general.

Re: Perspective
  • 9/24/2017 8:23:14 PM
NO RATINGS

Kq4ym, there may be pure uses where automated analytics can deliver a human less solution and be accurate, but my guess is that the pendulum will swing back as even the diehards realize that humans in the very least must confirm what the automation is determining.

Re: Perspective
  • 9/24/2017 4:27:16 PM
NO RATINGS

It does seem true that we might consider data science to " augment human work and decision making," but there are those folks who want to go further and automate, maybe not only for cost savings to companies but for those folks with lots of curiosity and inventiveness, becasue it can be done.

Re: Perspective
  • 9/20/2017 8:31:39 PM
NO RATINGS

..

Rbaz writes

Augmentation is the theme, but we do flirt with the possibilities of achieving full automation. Pipe dream, hype or wild imagination keeps it in the conversation. Who knows, maybe we may be rethinking this in the not too distant future.

That was basically my own reaction to the report in this blog that ...

Ultimately, all of the panelists agreed that the goal is to augment human work and decision making, not necessarily to automate it all.

"There is a great difference between automation and autonomy," Schabenberger said.

A little voice in my head kept saying "Wish that were true ... but there's a battalion of very nerdy people out there working away assiduously and obsessively to try to ensure that absolutely everything does get automated ..."

I also wonder what will happen to all those nerdy people if it does ...

..

Re: Perspective
  • 9/20/2017 4:18:02 PM
NO RATINGS

Augmentation is the theme, but we do flirt with the possibilities of achieving full automation. Pipe dream, hype or wild imagination keeps it in the conversation. Who knows, maybe we may be rethinking this in the not too distant future.

Perspective
  • 9/20/2017 1:40:43 PM
NO RATINGS

I like the "augmentation" theme. I've been hearing it more and more. Maybe it will help too calm all those fears about AI booting everyone out of their jobs.

INFORMATION RESOURCES
ANALYTICS IN ACTION
CARTERTOONS
VIEW ALL +
QUICK POLL
VIEW ALL +