If you strip away the buzzwords, “decision intelligence” (DI) is simply the discipline of turning information into better action — consistently, transparently and at scale. That’s why accountants should be leading it. 

As a profession, we sit at the intersection of information (business intelligence, performance metrics and forecasts) and risk assessment (likelihood, mitigation strategies and consequences). We’re trained to ask, “What could go wrong?” and “What evidence supports this choice?” That blend of insight and prudence is exactly what DI requires, especially now that leaders are leaning on artificial intelligence (AI) to generate options faster than they can evaluate them. 

A recent WorkLab podcast featuring South African data scientist and statistician Cassie Kozyrkov — who created the discipline of decision intelligence — offers timely reminders for any firm or finance team building AI-enabled decision workflows. Below I connect her core ideas to accounting practice and outline a practical DI playbook you can deploy right away. 

What DI Is (and Isn’t) 

Kozyrkov defines DI as “the discipline of turning information to better action — any scale, any setting.” It breaks down silos between data, psychology, managerial science and risk. Put differently, DI is not “more dashboards.” It’s a repeatable decision system that starts with a clearly defined question, pre-commits to how evidence will be used and specifies how uncertainty and risk will be handled before anyone looks at the numbers. 

That sequence matters. As Kozyrkov cautions, many “data-driven” decisions are actually data-decorated — we pick numbers to justify a choice we’ve already made. Her antidote is to set the goalposts before you kick the ball: pre-define the metrics, thresholds and tiebreakers that will determine the choice. That’s objective logic applied to decisions, not just transactions — squarely in accountants’ wheelhouse. 

Why Accountants Should Lead DI 

1. Evidence discipline: Our profession formalizes documentation, materiality and reliability of information. DI needs the same rigor: a decision charter that records the question, criteria, acceptable risk and returns, and sign-offs. 

2. Risk fluency: DI requires surfacing trade-offs and designing guardrails. We are trained to quantify likelihood, impact and control design — not just expected value. 

3. Ethics and accountability: DI is as much about governance as math. We bring independence, objectivity and transparency to how decisions are made and monitored. 

Pair that with AI’s ability to generate options and first drafts, and you get what I call the ultimate leadership partner: human judgment amplified by machine creativity, bounded by well-designed controls.

Avoid Pitfalls in the AI Era 

The two big DI pitfalls include the following:

1. Too many options and analysis paralysis: Generative AI can produce thousands of plausible answers. More options do not automatically yield better decisions. Without option limits and tie-breaker rules, teams waste time comparing immaterial differences that don’t change outcomes (in the podcast, this is illustrated by the Paris vs. Madrid syndrome). 

   What to do: 

   • Cap option counts (e.g., “AI may return 10 candidates; the team will use its criteria to short list the top three to evaluate in detail.”). 
   • Define tie-breakers in advance (e.g., “If two options score within two points, lower compliance risk is the more important criteria.”). 
   • Set a decision deadline (e.g., “Decision within five business days once evidence gathering is complete.”). 
2. Biased evidence gathering and selection analysis: Confirmation bias creeps in when teams look at data and evidence after forming a preference. As Kozyrkov notes, you can always move the goalposts post-hoc.

What to do: 

• Define the criteria before gathering the evidence or looking at data around options.  
•  Separate roles. The person gathering the evidence shouldn’t know the weighting of the criteria to prevent handpicking evidence or presenting evidence in a way that makes a preferred option look better. 
• Log deviations from the selection plan and require stakeholder acknowledgment. This creates transparency around the evaluation of evidence.
  

The Mindset Shift

AI changes the pace and shape of choices, but it doesn’t change who should be in the driver’s seat. As Kozyrkov puts it you — the human — are the “author of meaning.” In firms and finance teams, that author is often the accountant who can translate strategy into criteria, evidence into action and risk into resilient outcomes — all in a way that is unbiased and objective.

DI needs exactly that kind of leadership. Our profession is built for it.