3 questions to ask to avoid a common AI pitfall

What determines success and failure with internal AI projects?

Recently, I was given an opportunity to sit down with executives at Dell Technologies during their Analyst Summit to better understand the state of AI infrastructure deployments within enterprise organizations. Dell Technologies has more than 3,000 enterprise customers that have implemented its AI Factory portfolio -- a collection of software, infrastructure and services from Dell as well as key partners, such as Nvidia. Dell is helping organizations stand up their internal AI initiatives using their own private or sovereign data.

On the topic of what factor plays the largest role in determining success with AI, multiple Dell Technologies executives highlighted that the difference comes down to leadership. "Organizations will not spontaneously improve themselves," according to Michael Dell, founder, chairman and CEO of Dell Technologies, who highlighted that success in AI must come from the top down.

Yes, the infrastructure, the data and the tools all matter, but if the organization's leadership isn't convinced of AI's potential, the likelihood of success dwindles quickly. For employees, there's a wealth of uncertainty surrounding AI, both for good and for bad. For every employee who sees AI as an opportunity for a productivity boost, there is likely another with concerns over job security.

Buy-in and commitment are essential, but what specific qualities of leadership make the difference? Further conversations revealed that organizational leaders who focus on the specifics of the process and use case that the AI initiative is expected to improve tend to have a better chance of success than leaders who are more focused on the specific tools being used.

Again, the right technology, tools and data are important. However, selecting the right ingredients isn't enough to ensure success. Leaders must also focus on the specifics of the AI use cases, the specific project goals -- both strategic and tactical -- and the key measures of success.

In other words, privatizing an off-the-shelf model and asking your employees to experiment and identify use cases is likely not an optimal path to success. In addition to the uncertainty that surrounds AI projects, there is a strong possibility of overinflated expectations. If your employees expect AI to deliver on a particular use case 100% of the time, and it delivers 80% of the time, then it can quickly be deemed a failure and fall into disuse.

Instead, leadership teams must be involved in identifying specific use cases, specific processes and specific goals. For example, identify an existing sales quoting process that requires 12 steps and actions from four individuals, and aim to use AI to reduce the number of steps to three and the number of people involved to two for 80% to 90% of cases. This is a rudimentary example of the necessary level of process focus, but it highlights the type of detail required when defining use cases and measuring success. Again, the infrastructure, tools and data are important, but if you can't define the use case to this level of detail, your project is already at a disadvantage.

3 questions to guide successful AI initiatives

For IT and business leaders, addressing these three simple questions will help better position your organization for success in AI.

1. Are you committed to AI success? AI isn't something you can simply delegate and expect to achieve success. As with any business initiative, AI projects need clear objectives and clear measures for success. Those measures must be implemented from the top down, from the leadership team, on both the business and technical sides. With any emerging technology, there will be unforeseen hurdles and new lessons to learn. It's essential for the leadership team to remain committed to success and be willing to pivot when necessary to ensure that the team stays true to delivering the vision that provides the highest value to the organization.
2. Do you know the process specifics? A lack of understanding of the process being improved and having unrealistic goals can doom an AI project from the start. Misalignment between expectations and results can hinder future AI projects. Leaders also need to keep the business focused on which specific AI projects get prioritized. AI should be deployed to assist the organization's most important people in doing the organization's most important work. Internal AI initiatives can be hindered by too many competing priorities and use cases. The leadership team must deliver clear and defined priorities to ensure focus within the organization, which will likely include saying no to many good ideas early on.
3. Are you operating at the right pace? Time to value is far more important with AI projects than it is with other digital initiatives. AI is still a rapidly evolving field. The level of technology available in two years will likely be very different from what's currently available. For infrastructure, that means deploying flexible designs that allow for the integration of new tools and technology. In addition, AI projects must deliver rapid value, measured in only two to three months, not years. If it takes three years to measure a meaningful return on an AI initiative, the technology will be outdated before the project has delivered significant value. As a result, when selecting use cases, it's important to identify smaller, focused objectives that can provide value quickly.

There are myriad factors that determine success with AI initiatives. But if the AI initiative doesn't have the full support of leadership and the use cases aren't clearly defined, the other factors will likely matter little.

Dell Technologies works with multiple partners, such as Nvidia and AMD, on its AI Factory platform, which can span both data center and hybrid cloud environments and is augmented with a complementary portfolio of services and software. Dell isn't alone, however. AI Factory or infrastructure platforms are also available from multiple other IT infrastructure providers, such as Cisco, HPE, IBM, Lenovo and Nvidia. In addition, public cloud providers such as AWS, Microsoft Azure and Google Cloud offer AI infrastructure services as well.

Given the importance of use-case selection and how use-case specifics determine data selection and preparation, the addition of vendor or partner services to help define and refine the use case will likely offer significant benefits, regardless of where an organization deploys its AI initiatives.

Scott Sinclair is practice director with Omdia, covering the storage industry.

Omdia is a division of Informa TechTarget. Its analysts have business relationships with technology vendors.