June 6, 2021
AI initiatives always start with the best of intentions. You create your AI strategy. You get the budget approved. You find some qualified data scientists to do the work. You identify the first AI project and have your kickoff meeting with the business partners. Your data scientists finally get the data they need. After a few weeks of data analysis and feature engineering, they build a few models. And JACKPOT! The models work!
At long last, your hard work and determination are paying off. Your data scientists present the model results to the executive steering committee. Everyone loves the progress and is excited to see your team leading the company into the AI future. You make plans to improve the model and create the engineering infrastructure to deploy it.
That’s when the problems begin.
Your data scientists start to complain about lack of input from their business partners. They have questions about improving the model and making precision/recall tradeoffs. Progress requires timely and detailed feedback, but the business partners are providing responses like this:
Weeks tick by with little progress. You set up meetings with leadership to try to break through the roadblocks. But rather than moving forward, you find yourself in a loop, answering foundational questions you thought you had already laid to rest. You’re rehashing questions about the project’s goals instead of pursuing those goals.
And then suddenly it hits you: your business partners don’t share your enthusiasm for AI. They’re not challenging the project on its merits; they’re giving you the “slow no” by passively resisting change.
This outcome is all too common. Many data science projects die a slow, painful death because the organization isn’t motivated to make it succeed. The failure arises for three primary reasons:
Very few people in your organization will easily understand the difference between a project based on machine learning techniques and one based on traditional rules-based software engineering. I’ve been teaching AI basics for years, and I’m always surprised at how few businesspeople understand the fundamental concepts even after a day-long workshop.
A one-hour kickoff meeting with business partners isn’t sufficient. Most likely they will walk out of the meeting without understanding the technology or what is required of them. Weekly meetings in the weeds of the model development process won’t cut it either. Your business partners are likely tuned out during these meetings, multitasking on other work. Chances are, they don’t comprehend the challenges your data scientists are tackling. Consequently, they lack a sense of urgency and an understanding of the scope of cooperation necessary to make the project succeed.
In some cases current managers might not like the implications of AI. The investment in data scientists, tools, and infrastructure for AI might need to be offset by budget cuts to IT and lines of business that matter to your partners. Sometimes projects that they have been working on for years will be either defunded or made worthless because of AI.
For example, new AI initiatives might make companies question the value of further investment in rules engines, workflow applications to support current business processes, and robotic process automation (RPA). The teams that have been working on these projects won’t be excited about competition from AI solutions.
Sometimes the resistance results from fear of change. AI modernization requires companies to rethink entire business processes. People like the idea of innovation in the abstract, but they might reject specific changes that seem to conflict with their worldview. Resistance grows when change requires a lot of work or threatens to upend a comfortable work environment.
Fear, self-protection, and misunderstanding can work together or alone to transform your initially engaged business partner into a reluctant partner. To avoid the failure of your AI project, you’ll need to take action.
The best way to avoid the “slow no” from your business partners is to share accountability for the project from the outset. Here are some tips:
These arrangements won’t guarantee project success. But they’ll position you well to get support from your business partners. This support will be critical for getting feedback on model results, help labeling training data, or suggestions on the best interfaces for the solution.
August 15, 2021
A company’s transition to AI is incredibly hard. As non-tech companies look for ways to evolve their existing teams and software infrastructure to support machine learning, they often make a common mistake: the “just a binary” machine learning (ML) antipattern. This approach to deploying models considers the ML model as an isolated binary inside the existing infrastructure. Although seemingly reasonable, the antipattern is fraught with hidden dangers.
June 6, 2021
Many data science projects die a slow, painful death because the organization isn’t motivated to make it succeed. In this post we address the three primary reasons projects fail and provide suggestions for what you can do to overcome these challenges.:
May 2, 2021
Your goal as an AI leader is to get your teams to think like pros. You want them to strategically look for ways in which AI can lift the entire business instead of just solving a narrowly defined problem. Your team should constantly seek ways to advance the bigger vision of becoming an AI-driven company. In this issue of FeedForward, I’ll describe the difference between how pros and amateurs think about AI.