Intelligent Automation
Transformation in a Time of Uncertainty
November 8, 2023 | Seattle
Companies large and small find themselves unsure how to harness the power of artificial intelligence to benefit their businesses. The Bloomberg Intelligent Automation Roadshow made a stop in Seattle to take a deep dive into how organizations can implement AI and intelligent automation systems to enhance observable, predictable operational efficiencies and help create a path forward during a time of increasingly rapid technological evolution.
Speakers:
- Madhu Kochar, Vice President, Product Development, IBM Automation
- Durga Malladi, Senior Vice President & General Manager, Technology Planning & Edge Solutions, Qualcomm Technologies
- Gayatri Narayan, Senior Vice President, Digital Products & Services, PepsiCo
- Shri Santhanam, Executive Vice President, Analytics & AI, Experian
- Brad Surak, Vice President, Digital Aviation Solutions, Boeing Global Services
- Sara Vaezy, Executive Vice President, Chief Strategy & Digital Officer, Providence
- Taylan Yildirim, Vice President & Head of Cloud Software & Services Operations, Ericsson
Bloomberg participants:
- Edward Adams, Editor, Bloomberg Media
- Anna Edgerton, Seattle Bureau Chief, Bloomberg
- Lisa Mateo, Business Correspondent, Bloomberg
Event Highlights:
Opening Remarks
The productivity gap is the current hurdle for business, said Madhu Kochar, Vice President, Product Development, IBM Automation. “Automation has always been a force multiplier. We are at another historical moment now when Generative AI and automation is coming together, which is going to really help us think about how work gets done differently.” A recent IBM study revealed “amazing” possibilities and many use cases, with a focus on IT automation to get that work done.
Panel Discussion: Putting Artificial Intelligence to Work
With about 50 hospitals and 1,000 clinics, Providence is a very complex environment that has been using “classic” AI for 20 years, for things like fall and sepsis detection and clinical trial enrollment, according to Sara Vaezy, Executive Vice President, Chief Strategy & Digital Officer. Efforts have been amped up across four domains, and include optimizing where clinical and business aspects intersect. “Our doctors and our nurses are the biggest opportunity area given some of the limitations around the existing legacy technology infrastructure of healthcare. One of the biggest barriers to innovation in terms of how data flows is around documentation.” She spoke about caregivers typing in everything while with a patient, and then “pajama time,” when they are charting at night. “We looked in our data and can see a spike from 7pm to midnight, every single night.”
Bringing AI to consumer devices – cell phones, laptops, cars and more – is a very doable goal for Qualcomm Technologies, according to Durga Malladi, Senior Vice President & General Manager, Technology Planning & Edge Solutions. He spoke to the ‘why’ of that mission. “If you truly want to scale the full potential of what you can do with Generative AI in the cloud, it becomes cost prohibitive and it’s not very energy efficient.” It’s about immediacy. Every millisecond counts. It’s also about privacy and security; another reason to bypass the cloud. The obvious question is, “Can we seriously run these very large models on devices? The answer is absolutely, yes.” He offered the example of making an airline reservation by dictating it to his phone, which converts it to text so the relevant information can be extracted to complete the task. All the while, there is learning. “Two years from now our digital assistants will behave very differently. Maybe it knows that I like one airline versus the other and it’s going to retrieve that information. Personalization is absolutely critical.”
Shri Santhanam, Executive Vice President, Analytics & AI, Experian noted their focus is on regulation and safety, so large models and generative AI, in particular, are still being explored. “How do you manage them? How do you prevent them from hallucinating? It’s quite interesting to think about the governance and framework around these things. It seems to resemble more the framework for humans rather than, historically, the frameworks you put for models.” A technique they have started using for large language models is retrieval augmented generation. Rather than injecting the information into the model, it allows a careful, curated way to manage personal information and to use it to create use cases.
IBM Sponsor Spotlight: Unleashing AI for Business: How Businesses Can Use Automation to Accelerate AI Efforts
An IBM survey of more than 20,000 executives in 11 industries in 21 countries shows a 72% growth rate they attribute to adopting Generative AI. It’s their competitive edge, said Madhu Kochar, Vice President, Product Development, IBM Automation. “During the pandemic, every business suddenly had to do a transformation that would have traditionally taken 10 to 12 years, and do it in one. If you didn’t do it, you probably don’t exist anymore. That was really the focus; to understand how they started this journey, because each of them is on a different journey.”
She believed another finding was an “Aha! moment,” until she listened to the prior panel that day to realize there is an overwhelming consensus that the benefits of using Generative AI outweigh the risks. Survey respondents were calling it “super-charging automation.” “The fundamental thing was that they believe that technology is going to lead, no matter which industry we talked to, no matter what type of roles they had. It was very clear that technologies will take them to the next level.”
In Conversation With: Gayatri Narayan, Senior Vice President, Digital Products & Services, PepsiCo
A massive, legacy company like Pepsico has to start its automation transformation from a distinct place. Gayatri Narayan, Senior Vice President, Digital Products & Services, said it starts with determining the lay of the land, considering disparate systems that scaled differently across various markets, and being realistic. “We just said we’d never be able to bring everyone up to the same level, so from there, we asked how we could harmonize this.” Their energy is focused on what they call the first mile and the last mile, with a domain-specific data template created in the business’s middle layers, harmonizing the business logic in the industrialized AI models. Intelligent automation and AI were applied as accelerators. “We empower local teams to use intelligent automation to plug into the broader unified infrastructure. The second part of it was around the evolution of business logic so everyone can keep up.” Last mile customization ranges from language translations to tax rules.
Panel Discussion: Advanced Automation and ROI
On devising a good ROI, Taylan Yildirim, Vice President & Head of Cloud Software & Services Operations, Ericsson, said his role includes executing programs efficiently. He offered an example of one where his team was responsible for testing software from around the world for one of their biggest customers. Each program worked fine on its own, but issues arose when the software was combined. It was becoming a time-consuming, expensive process. “So, we decided to invest in test automation, instead of spending three or four months validating everything in the software. You have to think about the key thing you should focus on, and invest in the right place. If I compare this program with other programs, or if I look at the ongoing program, which is more heavily invested on the automation part, you can see more margins and the real savings from the products.”
At Boeing Global Service, a lot goes into not just preserving jobs, but making employees more effective. Brad Surak, Vice President, Digital Aviation Solutions, said it’s not just about software engineers anymore. “For instance, we do a lot of maintenance on aircraft, and so we train those people in simulators, and we can actually put reinforcement learning into the simulator.” This enables an understanding of where individuals need more training, what procedures can be simplified or when they can start using techniques like large language models. “We can start to take these complex maintenance procedures and guide the repair in the moment. So, it’s not about replacing the person, it’s about helping really give that person better situational awareness, helping them make the best judgment on what’s the next best action.” The same goes for operational centers, where simulation includes not just aircraft but the airspace at specific airports. It’s about relieving the pressure with automated analysis.
This Bloomberg briefing was Proudly Sponsored By
——————————
Join the Conversation: #IntelligentAutomation
Instagram: @BloombergLive
LinkedIn: Bloomberg Live
Twitter: @BloombergLive
Interested in more Bloomberg Live virtual events? Sign up here to get alerts.
——————————