It’s day two of the Gartner Digital Marketing Conference, and “AI” is definitely the buzzword du jour. AI for marketing came up in every session I attended today – whether it was led by a Gartner analyst or a technology vendor.
Did you know that AI can detect sarcasm with 82% accuracy, but it CANNOT detect and eliminate bias from a data set?
This was one of the more provocative points made in the session called, “What AI Teaches Us About Being Human,” presented by Gartner’s Andrew Frank. According to Frank, the results of AI can be skewed by biases in the data because computers can’t detect biased data. That’s because they have nothing to compare it to; all they have is the data they are given.
Once I heard this, I immediately thought about what I said in yesterday’s post:
When [customer data] is so severely fragmented and inaccessible, it’s impossible to gain insights and take action in real time. Without a single view of the customer, AI for marketing will be an empty promise. Garbage in, garbage out.
After listening to Frank’s presentation, I want to double down on that idea. It’s not just about giving AI a single view of the customer, but rather, the most complete single view of the customer. That means you need to collect customer data from as many different data sources as possible, and you need to do it in an automated way if you want to reduce the risk of data biases influencing the output of AI.
It’s not enough to just look at the customer data captured on your website or just within your CRM. Marketers need to think about all the different sources of customer data that exist within their organization before they can truly achieve a single view of each customer that is also resistant to data biases. Only then can you confidently feed this view to AI for activation purposes.
To learn more about how our CDP uses out-of-the-box integrations to automate data collection from all your different customer data sources, check out this on-demand webinar: CDPs – A Marketer’s Customer Data Control Center.