Architecting the road to AI success: Reflecting on the Gartner Symposium Panel

Last Tuesday, I had the privilege of leading an insightful conversation at the Gartner IT Symposium with an incredible panel of AI technologists – Hillery Hunter, Hilary Mason and Margaret Dawson – about the obstacles that organizations need to overcome if they are to successfully implement AI. 

The foundation of our panel discussion was based on a study that we recently commissioned to understand how enterprise executives are approaching AI. I will share more of the research in coming weeks, but one big takeaway is that companies that see results from AI are keeping AI capabilities at the core of the business. While most organizations are still experimental, we are seeing a strong correlation between organizations with highly dedicated on-premise AI capabilities to high performance, measurable ROI and less failure. Net-net, on-prem AI capabilities and solutions are fulfilling the AI hype. 

But, let’s be honest with ourselves, building on-premise AI capabilities is a daunting task. Our panel conversation centered around the challenges that come with the strategic business choice to build and deploy AI in a controlled manner at the core. 

First, there are new rules for infrastructure.

With the research indicating on-premise AI capabilities lead to greater success, companies need to consider how they go beyond experiments run by data scientists with servers under their desks or in small labs. Our panel discussed how asynchronous processes has different compute demands than traditional workloads, which is why we see the need for a different type of datacenter approach – one that enables compute power, control and productivity. 

During the panel, Margaret Dawson described how the case for keeping AI capabilities core and controlled becomes stronger because, often times, new AI technologies need to tap into the transactional and historical data that sits at the heart of the organization in servers and mainframes. She reminded the audience that “the world is not binary – the world is hybrid when it comes to AI.” 

Second, collaboration is a challenge everywhere - there are new organizational and cultural requirements.

FOMO is outweighing pragmatic business rationale for AI transformation. We all agreed that it’s critical to not get caught up in everyone telling you to “do AI.” Many times, the benefit of building AI capabilities on-premise will be that the organization can start deciding when to NOT implement AI as a solution to a business challenge. 

Building the right collaboration model for AI transformation is not easy and what works for software development may not work for collaboration across the lines of business, IT and data scientists. 

According to Hilary Mason, “Agile is a beautiful philosophy, but it’s where data science goes to die.” And Hillery Hunter added, “Agile is not an excuse for not knowing where you are going.”

In previous articles I have mentioned how we are at Peak AI Hype and how when you combine an overabundance of hype with unbridled spend, organizations end up without a clear path to ROI. The quandary I continually find when speaking with executives is: when will hype and experimentation transform into widespread adoption with real ROI?

While technology is central to the conversation, I believe that leaders also need to start thinking about the success factors that go beyond the technology. I see three key elements of a successful AI transformation.

1.     Start with the core business challenge. Target business issues that have a potential positive impact that is broader than an isolated case example for experimentation.

2.     Organize for collaboration. Build a cross-functional team of experts that includes business leaders to IT to data scientists to ensure all stakeholders rally around the business impact and are culturally ready to embrace AI transformation. 

3.     Repeat and scale experiments. Document the expectations and processes of the center of excellence in order to ensure efficiency. Real, long-lasting results come with scale and repetition. 

What I have observed is that organizations focus almost exclusively on the data and technology elements of AI vs. focusing on the cultural and organizational transformation that AI will demand from organizations and people. Our panel was adamant that we need both.

Thank you again, Hillery, Hilary and Margaret for joining me on stage. It was rewarding to host such an incredible group of experts.  

About the study: We wanted to dig deeper into the DNA of successful AI thinking, design and delivery in major corporations, so we commissioned a study from Inc.digital whose managing partner, Michael Gale, is the author of the Wall Street Journal best-selling book on digital transformation, The Digital Helix. As part of the research, we interviewed 566 executives and departmental leaders in companies over 500 employees. The design was focused on answering six key questions around how they were building their AI engines for success.