<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=520757221678604&amp;ev=PageView&amp;noscript=1">
Aimee Pagano

By: Aimee Pagano on March 18th, 2024

Print/Save as PDF

Is your data a wonderland for genAI?

Data Protection | data integration | data activation | data centralization | data governance | AI for associations | GenAI | LLM

mailto:demo@example.com?Subject=HighRoad Solutions - interesting article
Hungry Hungry Hippo was a cool game. The strategy was pretty basic. You had four different hippos all chomping down on plastic marbles. Each player had a hippo. The goal was to hit the hungry hippo at just the right time to gobble up the marbles. But, as kids, that translated to hitting the hippo lever obsessively to get as much circular yumminess as possible. Those with the fullest hippos at the end of the clock, won the round. And so on and so forth. 
GenAI is a little similar. Yes, per usual, I’m oversimplifying. I’m comparing one of the most powerful and evolving technological advances in history, to a $15 Milton Bradley game from the late 70s. Yes, this is indeed happening.
But roll with me here. 
GenAI consumes. It feeds on data. The better the data—meaning cleaner, more precise, more current, more cataloged, more accessible—the better the hunting ground is for GenAI. 
As such, even if it’s just a raw idea at this point, any orgs looking to integrate org-wide generative AI and Learning Language Model (LLM) solutions that tie their business ops together, need to put the focus on their marbl— ahem…their data today.
For-profit orgs are already quickly finding that single AI-powered solutions just aren’t working because they’re only looking at fractions of the data and/or serving siloed slats of their business. They’re now installing consolidated enterprise-wide tools with multi-pronged functionality powering every corner of their business. 
But that’s just not enough. Why? Because they haven’t put the work into data centralization or data governance to meet their long-term or even short-term business goals.
Assocations and nonprofits are no different. I’m confident there are a number of professional associations, trade orgs, mission-based nonprofits, hybrids, and even user groups out there exploring nook and cranny AI for your orgs. And that’s great. Martech evolution, also great. Change when change is needed,  excellent.
Adapting solutions for efficacy and efficiency is power. But, if you’re working toward org-wide AI strategies within your org, you need to get your data, and the people who sustain and leverage it, in the right place. 
Data volume, data collection, and data recordation are nothing without system integrity and data activation. Holistic, AI solutions that add sustainable value are built on the backbone of mindful, intended data. 
Creating your wonderland
So, if you’re exploring org-wide AI solutions in the future (think even 1-3 years away), you want to get your data house in order now. Here's what you should be looking at: 
Data centralization & curation
Matching and organizing socks is a big challenge, particularly for big families. All too often, half of the socks are in another load in the dryer. The best, and only, way to create matches is to put all of your socks together. Only then can you find functional pairs.
That goes for data centralization as well. It order for your data to be fully functional, it has to be centralized. Otherwise, you're missing half of the picture. That doesn't mean it needs to literally exist in one space. The best centralization comes from consolidating tools into those that can handle multidisciplinary, org-wide functions, and integrating sources where necessary. This—along with external data sources—is your AI's playground. 
Data quality & hygiene
Getting all of the Legos off of the floor so that you don't step on them is huge, right? They're a distraction. They create a labrynth that you have to walk through. And wow it hurts when you step on them. Think of the same for GenAI. Coming to terms with the fact that your data may not be as hygienic as you'd like it, is the first step to achieving data integrity. Setting up ops workflows and SOPs that scrub and maintain cleanliness is critical. Here are just a few of the Legos you can clear out and clean up, all easily handled through automation, and in some cases, integration: 
  • Invalids
  • Dups
  • Opt-outs
  • Unpopulated properties
  • Unpopulated values
  • Unappended information
Data prompt standardization 
Standardizing data across systems is one thing. Standardizing the way people store, manage, and leverage that data is another. But with AI, standardization goes even further. It's not enough to standardize the way data is handled. The language that happens between the human and the machine needs to be systematized across the association.
Now, sitting alongside data catalogs and dictionaries, Data Libraries need to include AI Prompt Repositories. Through these repositories and SOPs, org staff can document and categorize prompt use cases as templates, so that they can be shared across the organization for consistency and efficiency. 
Data policy
This feels like a good time to bring it back to those Hungry Hippos. Picture four primary-colored hippos and four very enthusiastic children pulverizing the levers to scoop up the most marbles. There's always one that goes a little too far, and before you know it, the yellow hippo is broken—now completely out of commission. That one misstep impacts the whole game.
Look at AI usage the same way. With the onset and ongoing adaptation of AI, orgs are now tasked with updating their policies to address how and when to leverage data sources across Learning Language Models (LLMs) and open genAI tools. Which takes us to...
Data protection
Piggy-backing (no pun intended...seriously it just happened) on the hippo analogy above, your org's and your members' information needs to be protected. AI ingests data for its designed output. There's no bias to what it consumes unless parameters are put in place to protect it.
And frankly, AI is still new. Based on the way that it's evolving, it will stay "new" for quite some time. With new things come exposure and risk. So you just want to make sure that your policies and practices seal this gap.
Since AI tools need continuing education, make sure they're not leveraging your members' information as a training ground. That means, any time one of your staff members plugs proprietary and/or personally identifiable information (PII) into the system, it naturally—because that's the whole point of a system that learns—gets added to your AI's training curriculum. That data can become part of a much larger playground. Understanding this, along with data policies that have been around for a while (i.e. GDPR, CCPA, etc.) will help you shore up data protection. 
In the end, AI is fun. AI can be a huge support when leveraged the right way. But AI has no dietary restrictions or preferences when it comes to your information. As your most valuable currency, it's up to your organization to both optimize and protect your data, sustainably and intentionally. 

Data centralization means data activation 
As stewards of association data, we at HighRoad are committed to leveling orgs up on data activation. Learn how Spark  + HubSpot  can quickly put you in the results zone, for AI and beyond. Book a consultation today to learn more.

About Aimee Pagano

Aimee joins HighRoad Solution with 15+ years of integrated marketing and communications experience, primarily in client-facing roles within the association and SaaS space. Her specialties include persona development, content strategy/management, lead gen and awareness campaign development, and website development/optimization.