Within the next five years, every large enterprise will begin an AI transformation. After helping dozens of Fortune 1000 companies to modernize for AI, I have seen stark differences between those that succeed and those that fail. In this blog post I describe three common AI transformation patterns that don’t work. In part 2 I describe a pattern that does work.
Approaches that ensure failure
The root cause of AI transformation failure is attempting to make once-in-generation organization changes incrementally, using existing people and processes.
Failure pattern 1: Just build a data science team
The first failure pattern is most common among midsize companies that have small IT teams and limited technology infrastructure. These companies often have a handful of team members with the title of “data scientist” who spend most of their time analyzing data for marketing, creating dashboards, or performing traditional statistical tasks like actuarial modeling.
When the organization decides to begin the AI journey, they create an isolated data science team to take the lead. They usually task the team with “creating a model” to solve a business problem.
This approach never works. An AI transformation requires much bigger changes than simply creating models. The data scientists in these situations don’t get enough direction, feedback, or engineering support to succeed at their task. After a year or two, little gets done and the best data scientists quit.
Failure pattern 2: Rely on the existing technology structure
The second failure pattern has a promising beginning: the C-suite or board mandates an AI transformation. Unfortunately, the executive mandate isn’t enough.
Because bringing in new talent is an expensive prospect, leaders attempt the transformation by using the company’s existing structure. Teams get bigger budgets for AI projects and hire a few engineers.
This approach fails for the reasons most corporate innovations fail:
- The organization has the wrong technology workforce.
- When more immediate problems arise, AI takes a backseat.
- The teams attempt to manage risk by blindly following agile methodologies.
The involved teams have a lot of meetings and create roadmaps and plans. But they take very little action. What action does happen is largely incremental work on existing projects.
The best possible outcome in this case is that leadership recognizes the problem early and begins making changes. More often, key talent leaves the team for better opportunities, and as years tick by, the program never makes more than incremental improvements.
Failure pattern 3: Be the lone-wolf AI leader
AI leaders at most companies come from various backgrounds. Some work in innovation, some in tech. Others come from an existing line of business. Regardless, they all share a vision about the future of AI and its destiny to transform every business.
These leaders become the company’s AI evangelists. In failure pattern 3, management decides to provide the AI evangelists with some resources to begin the AI transformation but not enough to stand up a dedicated team. The AI leader hires a few team members and recruits part-time talent from within the organization.
Unfortunately the difficulty of the AI change far outstrips the available resources. The AI leader has to fight for the attention of the company’s most effective engineers. Product teams can’t help much because of existing commitments. Meanwhile, the rest of the company hasn’t bought into the change. The AI leader is sucked into endless meetings about topics like AI ethics and model governance before the team even deploys a solution.
The AI leader complains about the situation, but the company is averse to more investment until the team can demonstrate returns on the AI investment. Without dedicated teams and business-partner commitment, results never materialize. Key people leave, and ultimately the project shuts down.
Big change requires bold moves
In this post I described three common AI transformation failure patterns. As I explain in our book Become an AI Company in 90 Days, we are at the beginning of a fundamental technology shift that will change everything. AI will revolutionize businesses in our era just as the internet, computers, and electricity did in previous eras.
Attempting to begin the AI transformation by using your existing organizational structure will work no better than basing an electricity strategy on gas lamps. Big changes require bold moves. In part 2 I describe how you can get started.