The first principle from the GOVERN framework of "A Culture of Governance" by Morgan Templar
The dawn of “AI for Everyone” took most of us by surprise. AI wasn’t new - we have been building AI systems for more than 20 years. But everyday work with AI was limited to a select few with the requisite background in data science, mathematics, and programing knowledge.
Today, everyone can access AI. Literally dozens of new AI-powered products hit the market every week. The pace outstrips the bandwidth of even those people who spend their life following and tracking these innovations for their newsletters or podcasts.
As someone who has been openly critical of chatbots, my skepticism is rooted in the belief that I’ve already explored everything a chatbot could possibly offer. When I reach out for support, it’s because I need human insight for complex issues. However, "AI for Everyone" has started to shift this perspective. Some forward-thinking companies have revamped their chatbots with cutting-edge LLM-based AI, making it possible for me to conduct more extensive research independently and efficiently.
Problem number one solved - reduction of calls to customer service
Problem number two solved - happier customer
There is a HUGE caveat to this problem solving.......The DATA.
The evolution of AI presents a double-edged sword, with the most significant edge being data quality and governance. For AI to truly serve our needs, the data fueling it must be of high quality and well-governed. This ensures confidentiality where necessary and that the responses provided reflect an accurate understanding of reality.
In writing "A Culture of Governance," I addressed the need to nurture a data-centric culture within organizations, and "Get Governed" offers a practical roadmap to establish a robust data governance program. However, the journey doesn’t stop at establishment; it’s about ongoing growth, adapting to new technologies like AI, and ensuring your data governance program is not just surviving but thriving.
Data culture is the important fertilizer to keep a data governance program growing healthy and strong.
Image by ChatGPT-4 prompt, Dall-E 3, prompt by Morgan Templar 2024
AI for Everyone has suddenly made DATA the safe harbor in a storm. New importance is placed on governing data. Boards are suddenly clamoring for it, rather than being fairly insulated from it.
The GOVERN framework, introduced in my work, stands as a compass for governance professionals navigating the expansion of their data governance programs
G = Grow the Program
O = Optimize & Operationalize
V = Value the Data and Value the Program
E = Evolve the Program
R = Revisit Core Principles
N = Negate Naysayers and What’s Next
In the Grow chapter I cover three main ways to grow an existing governance program:
Identify critical data and make sure it’s governed
Govern new assets
Streamline systems and business processes by eliminating redundancies
There are three kinds of data:
Known Governed Data
Known ungoverned data: information that has been identified as not critical to include in a governance program
Unknown ungoverned data (or ‘dark data’)
First, let’s talk about KNOWN GOVERNED DATA.
Decisions made about what information to include in a data governance and data quality program have occurred over time. Even in the most carefully governed organizations, foresight into the current AI landscape was rare.
The processes and information needed today to authenticate, deliver, and utilize data at our fingertips is at an unprecedented rate. Yet, if we assume our small, most likely underfunded, data program can respond to this “crisis of availability,” we are fooling ourselves.
The requirements around Privacy, Data Ethics, Data Science Ethics, Data Observability, AI Ethics, and AI Governance are relatively new. They are certainly greatly expanded. In fact, I will state with certainty that AI Governance of the past, if it existed at all, will not meet today’s standards and requirements.
Second, let’s talk about KNOWN UNGOVERNED DATA.
Past decisions about what data and information needed to be governed were greatly influenced by system requirements with some emphasis on business process. Restrictions to only govern structured data have been common from IT groups. This leaves these business-owned assets in the ungoverned pool.
Some examples of this kind of data include:
SAS databases in a business shop.
Excel documents on shared drives used to make decision, report metrics, or guide processes.
Data in transaction systems may be collected according to standards, but is unlikely to be regularly evaluated or monitored.
Data in SaaS repositories, such as a CRM. The assumption is often made that the mastering done in a CRM is sufficient and doesn’t need to be governed - nothing could be further from the truth. At a bare minimum the Person Master ID must come from an internal MDM and fed to the CRM to avoid duplicate records.
This kind of data has the possibility of hiding great value if it were governed and available. It also may be harboring non-compliance or other Risks that haven’t been fully vetted.
Finally, let’s talk about UNKNOWN DATA.
This data is everywhere.
Shared drives, individual desktops, personal cheat-sheet in a binder, images, scanned documents with or without metadata, instant messages, meeting recordings, audio recordings, decisions made via telephone or e-meeting, minutes from meetings, email, printed records, and the list goes on and on.
By far, more of your data is in the Unknown Data category than in the Known Governed Data or Known Ungoverned Data.
The difference between Information Governance and eDiscovery is a Lawsuit.
If you are ever sued for anything, attorneys have access to all your data. eDiscovery tools can scan through unstructured data like water through sand. Everything becomes a potential liability or threat.
Inactive databases, which probably should have been deleted, but which are held “just in case we need it” are now part of the information that can be used against you.
Scanned documents are a particularly juicy treasure trove for eDiscovery. Their software typically converts everything to TIFF records (a kind of video format) and classifies, categorizes, and establishes metadata for huge databases of images. You likely have mounds of data that can show you are culpable for things you thought you had well governed.
Video and Audio transcripts can be used to show patterns of decision making. Have you considered what your typical decision-making paradigm means for your stated mission, values, and/or ethics policies?
Use these points as a starting place to identify what data and information in your organization need to be governed that aren’t. Also look at your Archive and Retention policies. Make an effort to clean up your Dark Data.
All things considered, a data governance program has a lot of ground to cover to truly ready an organization for frequent and pervasive use of AI. You can start small, use case by use case. Or you could tackle one system or database at a time. Examine strategic projects, such as a new CRM migration, as an opportunity to ensure governance is occurring and is embedded into the new processes and policies that will be written for this new system.
GROW your Data Governance program. Opportunities are plentiful. Start by identifying the area with either the most Value or the most Risk. Make a plan. Align the plan with the strategic goals. Get funding for the new initiatives. Set and report on Metrics and KPIs.
Stay tuned for more insights on fostering “A Culture of Governance” through our First CDO Partners website, on Substack, and LinkedIn articles. Our series of videos will further explore these critical growth strategies, guiding you through the development of a robust governance culture.