Organizations today are rushing to explore and deploy AI applications; perhaps yours is among them? If so, it’s critical not to overlook the one budget item that absolutely must be present but too often isn’t: governance!
Many teams allocate tens of thousands of dollars (or more) for AI-related computing resources, model development, and deployment infrastructure. But by treating governance as an afterthought, they set themselves up for disappointment at best and failure at worst.
Poor Governance is Not an Option
A recent study by Informatica reported that 43% of data leaders say data quality, completeness, and readiness are among the biggest obstacles preventing GenAI initiatives from moving out of the pilot stage. Experience tells us that the issues that arise involve the likes of compliance violations and privacy breaches – and the worst part is that these issues tend not to become evident until months after deployment, leading to major costs in fines and remediation.
Consider this: A $200,000 investment in governance could prevent a single privacy violation that might lead to literally millions of dollars in fines. (To this point, here’s a post that will make your blood run cold.) At the same time, solid up-front governance likely also will accelerate AI development by providing clean, well-documented data that reduces model training time and provides significantly higher-quality results. (See my thoughts on GIGO* for more.)
Positioning governance as a means to both reduce risk and unlock opportunities has been a mantra of mine for quite a long time (here’s some more about that), and AI budgeting is a perfect place make this point to your powers-that-be.
Building Governance Into Your AI Budget
When it comes to budgeting, please allow me to suggest the following as a rough guideline:
- Open the bidding at 20%: Allocate at least 20% of your AI project budget specifically for governance activities. At minimum, this should include ROT removal; deduplication; data classification; privacy, security, and ediscovery safeguards; and quality assurance.
- Fund governance roles early: Budget money and time analysts from day one for the involvement of records managers and information coordinators; data analysts and IT staff; and privacy and security. Outsource these functions if you have to because these aren’t optional support roles – they’re core team members who can keep you from making expensive mistakes.
- Invest in governance technology as needed: Include costs for such systems as data cataloging platforms, automated classification tools, and privacy impact assessment software. These tools can pay for themselves by reducing manual effort and ensuring consistency.
The takeaway here is to remember that information governance is a strategic enabler of any successful AI implementation (as well as so many other critical processes, of course). Including it in the AI budget therefore is not just appropriate, but central to the overall outcome.
*Garbage In, Garbage Out
What percentage of your AI budget goes to information governance? Share your thoughts below!