If you’ve seen the movie Titanic, you know what can happen when you venture into dark, uncharted waters without a true understanding of the risks.

By the time someone yells, “Iceberg, right ahead!” the danger can appear out of nowhere and your ship sinks as fast as the RMS Titanic did in 1912. 

The same is true of AI.  

The promise of AI rests on a foundation that can only be built if you avoid dangers that are lurking right under the surface.     

AI Dangers

Third-party AI services like Microsoft’s Azure OpenAI Service are attractive because developing AI apps can be enormously expensive. Rather than building models from scratch, businesses can use preexisting models and services and then scale them to their needs.

But using third-party services can also create data privacy and security concerns because to create AI outputs, services must store the inputs. And often the developers themselves do not have clear visibility on how the models work, how the data is managed, or if and how the data is stored and protected.  

As troubling for companies to consider, third-party AI vendors want to legally protect themselves in the event of problems, so liability could likely fall on the businesses who share the data to begin with.  

While the legal groundwork around AI is still being laid, the most important thing for organizations to do is to start upskilling their teams around AI immediately so those employees become your AI support system as your AI program moves forward.  

Bad data in = Bad data out 

Because the consequences of bad data on enterprise AI programs could be broad and vast, Enterprise AI must start with a solid understanding of the data that will power the AI models. 

That activity should take place agnostic of the technology downstream. It is a strategic exercise to identify how the data should be organized, ingested, and managed. 

Then and only then can you talk about how AI will use that data and make it actionable for the teams responsible for moving your company forward.

Identifying good use cases for AI should be a top priority for leaders, and that means thinking about your goals first and then analyzing how to get there through the lens of people, processes, and tools.

With AI, focusing on upskilling your teams so that they can be your internal ballast for the AI conversations to come over corporate alignment, risks, and risk mitigation.

How Data Cloud can help  

It’s clear that the data matters.  But enterprise data generally tends to be stored in multiple platforms that were brought into the organization for many different reasons. 

The marketing team likes Platform A because it’s easy to use. The POS team has been tethered to Platform B, a legacy system that works but it’s not very flexible making any changes incur high levels of effort. And the customer service team is working on Platform C, a cloud-based system that’s not connected to anything else making it difficult for the service teams to have timely relevant data to contextualize their conversations with customers. 

Take comfort in the fact that this isn’t just you... this is so many businesses.

The task at hand is to ensure your data is as clean as possible across these disparate systems and then rely on a tool like Salesforce Data Cloud to help centralize all data sources.

Data Cloud can ingest data from all platforms across your organization and model the data in a uniform manner for all data to “speak the same language.” Once it’s accurately modeled, you’re able to develop rules that unify the data creating unified customer profiles.

From here brands can utilize unified profiles and engagement data to populate models (including AI models) that help support business outcomes.

As an example: You’re looking to understand what actions lead to a sale and how to personalize the experience enabling the customer to convert quicker.

Instead of your teams manually pulling each data point from its data source to begin to answer this question, you’ll look to Data Cloud where it’s already centralized and unified to build a sophisticated model utilizing all your data in harmony.

But wait, how does this enable AI? 

All of this will allow your data to be prepped, modeled, and centralized to use in an AI model where the sophisticated, unified data will be analyzed based on the specified AI parameters. 

For this Predictive AI model your brand can pull historical data points showcasing actions that led to past sales, indicate who has been buying, showcases trends in the data and then create a model that identifies next best action for the customer earlier in the journey to result in a quicker conversion. 

Leveling up your customer experience through AI driven personalized customer experiences drives increases in your key conversion points.

Move over, Rose

Since the movie Titanic was released in 1997, we’ve all wondered why Rose and Jack didn’t try to both fit on the piece of the wooden door to stay out of the freezing water after the boat sank.  

Regardless of the physics, they didn’t even try to make it work.

With AI, you won’t have that luxury. 

With AI, customer safety and customer privacy must all sit on top of your data, which is the wooden door. 

Your challenge will be to make it all fit together and balance everything out so you don’t tilt, plunging into the cold, deep sea. 

Jennifer Johnson and Erica Skelly and work in digital strategy at Horizontal Digital. Erica is based in Chicago. Jennifer is based in Dallas.