Navigating AI Ventures with PRISM: Five Guidelines to Minimize Risk in Selecting AI Initiatives
Navigating AI Ventures with PRISM: Five Guidelines to Minimize Risk in Selecting AI Initiatives
Many initiatives, whether AI-based or not, frequently disregard fundamental design principles that significantly enhance a project's chances of success. Here are my five guidelines to avoid such oversights:
- Power Struggles
- Beat
- Organization's Persona
- Human Resource Management
- Analytical Measures
Some of these concepts may seem self-explanatory, yet I've witnessed numerous pitches for project funding that overlook these straightforward, fundamental concepts.
Power Struggles: Ensure the project's cost and reward fall under the same department head.
At the corporate level, all costs and benefits are overseen by a single administrator. However, as one ascends the organizational hierarchy, various departments and functions prioritize optimizing their domain before considering the rest of the organization. One senior consultant once shared with me, "As soon as I hit my sales target, I become a team player. If I meet my target six months ahead of the year, I'm a team player for six months; if I achieve it a month before the year ends, I'm a team player for a month." I've encountered numerous project funding proposals having the project's expense in one department, say, IT, and the reward in another, such as marketing. This is a warning sign for any project. Furthermore, there's an Academy of Management Journal article titled "On the Folly of Hoping for A While Rewarding B" that elucidates this issue.
Beat: Select projects with a shorter timeframe for benefit realization.
Areas like customer service, digital marketing, or trading operations typically have shorter cycles, making it easy to observe the project's impact. For instance, if your AI project improves a company's customer service representatives' efficiency or automates their tasks, the benefit can be noticed quickly. This is one reason why certain scientists focus on organisms with shorter lifespans, such as Mayflies, which live, reproduce, and die within 24 hours. I believe the best projects deliver some initial value within a year, and even within six months.
Organization's Persona: Connect the project to the company's core values.
Simon Sinek's "Start with Why" has gained popularity by emphasizing the importance of linking projects to an organization's identity. By connecting a project to your company's identity, it becomes easier to answer "why" this project is crucial. For example, after Steve Sinfosky, a Microsoft alumnus, wrote the "Computing at Cornell" memo, which kick-started a transformation at Microsoft, Gates subsequently penned his famous "Internet Tidal Wave" memo. This connection strengthens the purpose behind any project. For instance, in the insurance sector, employees can embrace the fact that insurance firms were one of the first AI-driven enterprises, as they have always competed on the merits of their decision-making models. This is a worthwhile association to make.
Human Resource Management: Invest in training staff on AI.
In his excellent book "C0-Intelligence," Ethan Mollick argues that the AI we have today is the least intelligent it will ever be. AI is constantly evolving and consistently becoming smarter. It is understandable that every organization and employee would benefit from having some familiarity with AI, whether as a user, chatbot creator, or even as a deep expert. Therefore, any investment in a project related to AI will serve as a valuable learning opportunity for staff, ensuring they are prepared for AI's future evolution.
Analytical Measures: Build on existing metrics instead of inventing new ones.
There is often a temptation to create new metrics for novel initiatives. However, developing new metrics can be as challenging, if not more, than implementing the project itself. Few organizations and functions change their metrics, making it essential to link any AI project to existing metrics. If your proposal includes a new AI project, making the linkage to existing measures straightforward and clear is crucial. For instance, if you are introducing a technology solution for a call center, focus on metrics such as calls per representative, call duration, one-call resolution, and customer satisfaction.
Questions to Ponder
When considering AI projects, keep these three principles in mind:
- How does my project rate on these three criteria?
- Can I clearly and convincingly connect my project to each criterion?
- Based on these standards, should I modify, cancel, or accelerate certain projects?
These questions can be asked early in the project development process and are challenging to address once the project has commenced. It's best to ask them boldly at the beginning rather than as a postmortem assessment.
In the context of AI projects, incorporating fundamental design principles can mitigate risks associated with generative AI implementation. For instance, clearly defining the department responsible for both the costs and benefits of an AI project can prevent internal power struggles, ensuring the project's success. Additionally, integrating AI projects with the organization's core values, such as decision-making models in the insurance sector, can strengthen their purpose and enhance employee engagement.