Trending Topics on AI in the Enterprise

We interviewed several AI experts in our Glass Tank at the O’Reilly AI Conference — here are the topics they are most passionate about.

Trending Topics on AI in the Enterprise

“We as humanity are having this awakening,” says Chris Duffey, head of artificial intelligence (AI) innovation and strategy at Adobe. “There is this technology out there that accelerates and augments our thinking. Framing it in that sense, the sky’s the limit.”

Observations like this and more were debated in the halls of O’Reilly’s AI Conference, where we had a chance to talk one-on-one with industry game changers about where AI is heading and what it means for the enterprise.

The big takeaway? AI is changing the way we work and simultaneously fueling the next enterprise digital transformation. Here are the highlights we overheard.

1. Work with AI, not against it

First things first — robots aren’t coming for all of us or all of our jobs. So, take a minute to shake off any lingering fears, and turn your attention toward the real trend: Humans and AI working together can maximize organization, efficiency, and success.

“Companies are trying to figure out how best to implement AI, and whether they should embrace it or fear it,” said Daya Nadamuni, senior manager, corporate strategy and competitive intelligence at Adobe. “First, they have to understand what it is, then they have to learn how to use it. We are in the early stages of AI. As the understanding and usage evolves, in another five years, I expect AI to be mainstream.”

“It’s not human versus machine,” said Daniel Raskin, chief marketing officer at Kinetica. “It’s human and machine coming together to complement.” And as that evolution continues, we’ll all be tasked with determining where new lines are drawn. “Should we know it’s a robot?” asked Susan Etlinger, an Altimeter Group analyst. “Should it be able to make decisions on our behalf that are different from decisions made in the past?”

While no one knows exactly how it’s all going to shake out, the general consensus is a working-in-tandem outlook — that together, machines and humans can accomplish even more than either can alone, without any lingering fear or threats.

“AI-driven computers can be good at task-specific problems,” said Chris Benson, chief scientist of AI and machine learning at Honeywell. “The place where humans will have a firm place of their own is in creativity, something we are really good at.”

2. Overcoming bias in AI

Another hot topic was the notion of continued innovation and the increasing concerns over bias in AI. While there’s no question that innate human bias exists, machines, it seemed, would be capable of overcoming these hurdles and delivering completely neutral analyses and deliveries. But, unfortunately, that isn’t entirely the case. “It’s not that the algorithm itself is biased. It’s just absorbing the bias that exists in the world,” Susan said.

Human bias can find its way into otherwise unfeeling machine-learning processes. How do we overcome it? “We have to be careful not to inadvertently disenfranchise people,” Susan said. “We have to be very intentional about the world we want to create in a digital space.”

While on its face it sounds like maintaining machine-learning integrity is a manageable issue, many felt that these biases cut far below the surface and reflect overarching problems with the technology and innovation industry as a whole. “The biggest problem in AI right now is that it’s being built by a group of homogenous people,” Olga Russakovsky, assistant professor of computer science at Princeton, said. “For example, only 10 percent are women, but we’re building this technology that is in theory supposed to represent everybody.”

According to the Pew Research Center, Olga’s assertions speak to greater industry issues. The percentage of women seeking STEM jobs has dropped from 32 percent to 25 percent since 1990, and minorities continue to make-up a small percentage of the workforce. While recognizing this seemingly built-in AI bias is important, without concrete and actionable solutions, it’s possible these hurdles will remain.

“Even if you’re not programming AI to be biased,” explained science journalist Matthew Hutson. “It might pick it up on the way. We need to diversify the engineers that are building the AI.” Until then, added Kathryn Hume, Integrate.ai vice president of product and strategy, we’re creating experiences for “well-represented audiences,” when we should be creating experiences for everyone.

3. Committing to the customer first

While we’re all moving at the speed of innovation, it’s nice to know some things never change. As with any other innovation, implementing AI in the enterprise means being decidedly customer-first in our collective and individual processes and workflows.

As Theodora Lau, founder of Unconventional Ventures, summed up, customer-centricity is essential to success in any landscape, including this AI moment in time. “You have to figure out where AI fits in with what you’re trying to do for your customers.” To do that, she explained, “Ask, ‘What are the problems I’m trying to solve, and how can I leverage technology as a means to an end to get there?’” And with more data than ever, said Anand Rao, PwC partner and global innovation lead, “You can start personalizing the experience for customers.”

Though customer-driven experiences have been front-and-center for decades, with rising consumer demands and a greater emphasis on one-on-one touchpoints, brands are being pushed to keep pace with increasingly lofty — and increasingly high-value — expectations. To address these challenges, Ashok Srivastava, Intuit senior vice president and chief data officer  explained the process his organization follows. “We always start with what the customer needs are,” he said. “We call it CDI — customer-driven innovation.”

The goal is simple. “We want to use AI and data to see how we can solve their problems and think about it all the way through to experience design. We think about things from the customer perspective, and then we build the technology to address that,” Anand said. It’s a mass-scale personalization approach that, done right, “means better revenue, better retention, better experience.”

4. Incorporating AI intelligently

Despite the palpable excitement and explosive growth in AI, participants and attendees were unified on one notion: while it’s tempting, AI still needs to be incorporated intelligently and responsibly.

“If you really want to move the needle,” said Susan, “start to try to optimize a part of the business that isn’t working as well.” Look for challenges that AI and machine learning can uniquely overcome, and start there. But simply layering in the technology because you can won’t yield the kind of results you’re looking for — now or in the future.

Where to start (or continue) when it comes to AI integration? “Don’t worry too much about AI,” added Anand. “Start with where you have the most pressing problems. Look at your existing organization. Almost every organization already has a transformation going on.” From there, explained David Kiron, executive editor for MIT Sloan Management Review, “You need to develop a plan. Take a step back and get educated. Get a basic understanding of what AI can do for you. Understand what data you have and what data you might need.”

At the end of the day, though, it’s our responsibility to “keep the right perspective,” explained Chris. “There is a lot of hype around AI and deep learning, but it is a specific toolbox for specific problems — it’s not magic. Take the practical and pragmatic approach.”

Learn more about the ideas and recommendations from Adobe Think Tank to see what the present and future of AI look like for enterprises and what you can do to prepare your organization.

Recommended Articles