Hats, Haircuts & Tattoos

Decision Making in AI, Business & Life

I’ve been thinking about decision-making a lot this season. In “Atomic Habits”, James Clear writes that the quality of our lives is all about the small, daily decisions we make. In the Enterprise Software space, AI has taken center stage in most conversations we’ve had in the past 18 months. The potential unlock it can deliver for decision-making, speed and efficiency is both thrilling and daunting. The decisions it helps us make can influence the direction of projects and, consequently, the overall path of the company.

Clear presents a thought-provoking framework for understanding decisions by categorizing them based on their permanence and impact. Let's explore how this framework can be applied to decisions in the realm of AI.

Hats: Temporary and Reversible Decisions

“Hats" represent decisions that are temporary and easily reversible. When it comes to AI, these are the decisions you can change without significant consequences. Examples include:

-Job Description Creation: Using AI tools to generate job descriptions can be easily modified or replaced if they don’t meet expectations. These descriptions can be iteratively improved without long-term commitments.

-Communication Creation: Using AI tools to help create or re-word general communication / email templates to ensure content is clear, succinct and grammatically correct is a great way to utilize GenAI with a relatively low risk of impact.

-Knowledge Base Documentation: Implementing AI to document and update knowledge bases is another “hat” decision. The documentation can be continuously refined and the ability to quickly document processes for knowledge transfer is optimal over than the alternative of no documentation.

-Learning and Skill Recommendations: AI-driven recommendations for employee learning and skill development can be adjusted as needed, allowing for flexibility in addressing evolving needs and goals. The risk for an employee taking additional classes or adding new skills is a relatively low one.

- Adjusting Parameters: Tweaking the parameters of your AI models, like learning rates or batch sizes, falls into this category. These decisions are important for optimization but can be adjusted or reversed with relative ease.

Haircuts: Medium-Term Decisions

"Haircuts" represent decisions that are more impactful and take longer to change but are not permanent. In Human Resource (HR) tech projects, these might include:

-Learning and Skill Recommendations: AI-driven recommendations for employee learning and skill development can be adjusted as needed, allowing for flexibility in addressing evolving needs and goals.

-Performance Review Creation: Implementing AI to assist with creating performance reviews is a significant decision that can be modified, but changes require careful consideration and effort to re-align the system and processes. There is potential direct impact to employees for performance reviews which are inaccurate or done with a lack of personalization and care.

-Succession Planning: Utilizing AI for succession planning involves more long-term strategic alignment and impact, making it a medium-term decision. Predictive analytics, talent modeling and reporting can be very helpful to gain a bird’s eye view of your workforce and plan for the future. However, the impact to flexing toward AI-directed decisions which are based upon inaccurate algorithms / data training or bias can have a long-term, negative impact on the future of your company.

Tattoos: Long-Term and Irreversible Decisions

Finally, "tattoos" symbolize decisions that are long-lasting and difficult to change. In AI, these are the foundational choices that have far-reaching implications:

- Platform and Tools Selection: Deciding on your vendor is a big commitment in time, money and people resources. Switching platforms later can be done, but it requires substantial effort, budget and adjustment. Going with a vendor who opens your company up to potential litigation or security risks can be fatal to your company and career, so vetting your vendor carefully is of huge importance.

- Data Infrastructure and Architecture: Designing the underlying data infrastructure and architecture is a critical decision with long-term implications. Once established, changing the core architecture can be incredibly challenging and costly. The quality of data is a huge deal as well. Train your AI models using bad data—and your output will also be bad.

- Ethical Guidelines and Governance: Establishing ethical guidelines and governance frameworks for AI usage is a foundational decision (as well as ensuring your chosen vendor has a solid governance & security structure in place). These principles guide the responsible development and deployment of AI and are not easily altered once they are in place.

Making Informed AI Decisions

Understanding the nature of the decisions you need to make in Enterprise Software projects helps in managing risks and setting realistic expectations. By categorizing decisions (low, medium or high risk/impact) you can better assess their impact and permanence, ensuring that you approach each choice with the appropriate level of consideration. Just as with personal choices about appearance, tech decisions require a balance of creativity, pragmatism, and foresight.

?? Are you trying to figure out how to utilize AI or AI tools in your Enterprise Software / HCM project? Raven Intel can recommend some fantastic consulting firms and vendor guides to help you think through your decisions.