Brad Reynolds, Senior Vice President – Artificial Intelligence at Expedient speaks with Robert Lutton, VP, Sandhill Consultants, and Editorial Board Vice Chair, CDO Magazine, in a video interview about, the challenges with scaling AI, the need to portray quick small wins, addressing AI challenges as an organization, and how it is all about data in the end.
Speaking of the burning challenges in terms of scaling AI for organizations, Reynolds refers to the challenge as a flywheel issue. He adds that the scaling AI conversations start with big sets of transformative use cases and the problem starts because nobody understands AI enough.
Adding further, Reynolds says biting off a massive use case brings out issues because people hardly know the risk factors involved or the probabilities of succeeding. Therefore, he insists on focusing on smaller and quicker successes such as getting an enterprise chat tool so that employees can learn about the AI capability in a less controlled but capable environment. The next step, he says, is to figure out the light use cases where AI could be combined with internal data to help solve problems.
Sharing an instance of such a use case, Reynolds mentions coming up with root cause analysis (RCA) when a multi-city and multi-client outrage happens. He adds that a human doing RCA takes many hours and does not include all the data. Further, once the issue is resolved, the affected clients must be demonstrated well to understand what is going on.
In light of this, Expedient was able to combine its private AI service with AI models that run on the company’s hardware. The root cause analysis is written based on tickets collected from the client data that exists inside the organization's tracking system.
Adding on, Reynolds says that once the RCA comes out, the humans read and audit and the organization is ready to distribute it. He stresses not just being AI service providers but also AI consumers which requires justifying the power of AI as a team while bubbling up to the management.
Reynolds states that putting up little use case demonstrations not only helps organizations to experience it better but also portrays the ROI analysis at an executive level. The biggest issue with scaling, he says, is jumping far ahead with transformational AI use cases without doing blocking and tackling.
When asked how Expedient would address AI challenges, Reynolds says that if hired as a consultant, he would first ask the organizations whether or not they want to be AI-empowered. Then, based on the answer, he would list out buffet items to sample.
Elaborating, Reynolds stresses performing an ROI analysis and implementing lightweight initiatives to get the flywheel spinning up. He sheds light on the need for an AI policy, as it is imperative to unleash enterprise chat. Also, he recommends confirming the availability of AI-enabled tools, and assessing if they are used publicly or privately.
It boils down to a data conversation outside of chat, says Reynolds, about the aspects of data privacy, and security. However, not all data need to be secure, he asserts, as some data can be used with public models because it's not proprietary.
As Expedient explores, it focuses on particular use cases, the business value, and how the organization can help by assembling the pieces and bringing AI to the data.
Referring to a Gartner report, Reynolds affirms that although 80% of enterprise data exists on-prem, Expedient does not ask organizations to take that data and create a three-year data plan to move into the cloud.
Rather, Expedient takes the same AI capability and places it next to existing data, because AI models can pick needles out of a haystack, says Reynolds. He concludes that everything comes down to where the data is and how it can be accessed and scaled by leveraging AI.
CDO Magazine appreciates Brad Reynolds for sharing his invaluable data insights with our global community.
