In which Jill watches AI give data a little nudge to wake it up.

Those of us who began our careers in data warehousing were at the forefront of the “data as an asset” conversation. These days it’s become a drinking game: Every time someone says “Data is an asset,” you have to take a sip of beer, and every time someone says, “Data is the new oil,” you have to down a shot of tequila.

But it’s true. We were there before data was a thing. So while many of us will relate to Jennifer, others will read her question and wonder who the hell would be bored with the hottest trend in AI?

Dear Jill:

I work in marketing for a global manufacturer. Before taking on my role as Senior  Director of Marketing Analytics I worked on my company’s enterprise data warehouse team, moving through the ranks from coder to data loader to data steward to heading up the enterprise data management team.

My tenure as Director of Data Management taught me a lot. When we couldn’t find a global metadata management system that met our needs, we built our own. Ditto with master data management. We’d automated data prep and stewardship functions before they’d become commercialized, and I’m proud of that. Our success had a lot to do with my promotion into Marketing.

My problem is that people are asking about data again, and I’m expected to lead the way. Again. I thought data had become a commodity, which is why I was happy to change focus to marketing analytics. After all this time, do I really need to turn my attention back to data?

— Jennifer, currently in Europe

Wow, Jennifer! You’ve done a ton in the data realm! How is it we haven’t met yet? We should be best friends!  I get it. We’ve been around data for so long, it seems so 2005. But let me tell you why it’s so 2025 instead! One word: AI. (I guess that’s two words, huh?)

Artificial intelligence has hit the mainstream in a big way. Executives like Amazon’s Jeff Bezos, Facebook’s Sheryl Sandberg, and Cisco CEO Chuck Robbins have all represented AI as key to their corporate strategies. Robbins, speaking on CNBC recently, explained how AI and machine learning were the centerpieces of his company’s new “highly intuitive network”:

We’re ushering in a new era of networking that’s powered by intent, informed by context, and over time continues to adapt and learn and actually becomes intuitive, and is connected directly to what our customers’ business outcomes are.

Robbins went on to describe how a bank would deploy the intuitive network to deepen a high-value customer’s experience through “intent-based infrastructure,” with machine learning continually tuning responses to optimize customer experiences, thus obviating “hundreds of engineers” and “lines of code throughout the infrastructure.”

Whoa! Happier customers? Talent efficiencies? Competitive advantage? How do they do it?

Two words: With data.

AI and machine learning are driving innovations in chip technologies, high-performance computing, and cloud architectures. Like chocolate chips in the cookie batter, data is baked into all these innovations. It’s an assumed ingredient. But data skills are not.

When it comes to AI, data can’t be an afterthought. Whether we’re making an online purchase or posting on social media, every click—and increasingly, every turn of the doorknob, the ignition key, the Nest thermostat—generates a transaction that can be used to train a model or communicate a message. Orthogonal data from governments or third-parties further enrich this data, increasing its value apace.

New data sets are part of many lauded AI efforts. IBM’s Watson Health, for instance, exploits data from sources as diverse as drugstore chain CVS, consumer products giant Johnson & Johnson, and device manufacturer Medtronic. We data geeks understand that ingesting petabytes of heterogeneous data into a cloud platform isn’t a trivial thing.

Using “straight through processing,” increasingly smart algorithms can choose the best data source and pinpoint the necessary data. Models will be able to cherry-pick their own data and train themselves, leaving data scientists to determine how and when to apply the results. Likewise, data acquisition processes, aided by machine learning, can “guide” the data to the consuming application or knowledge worker at the time of the request. No more poring through different databases or emailing colleagues extract requests. The data can find the application at the time of consumption.

But an understanding of what it has taken and what it takes to prepare, cleanse, annotate and tag, semantically reconcile, and enrich data still matters. I’d argue it matters more than ever.

In fact getting the data out of platforms and into models might be the easy part. Sourcing the data in the first place is the harder nut to crack, as more global data standards and authoritative formats continue to be a cri de coeur for battle-weary data professionals. The new crop of data prep and wrangling tools that use AI rely on human “curators” to guide and tune their rule sets. Data-savvy professionals are integral to optimizing these processes.

My advice? Don’t give up on data. Assuming you can’t roll up your own sleeves, engage with your old colleagues to launch some data discovery projects, commission external data, and test out machine learning on existing data to reveal what’s next.

The data world won’t stop evolving. But these days, there’s more demand than supply when it comes to data knowledge. Two words: Share yours.

Two words: Share yours.

Original post on “Q&A with Jill Dyché” column on upside.tdwi.org