Artificial intelligence: a reality check

Artificial intelligence (AI) is the new black, the shiny new thing, the answer to all marketers’ prayers, and the end of creativity. The recent emergence of AI from the arcane halls of academia and the back rooms of data science has been fueled by stories of drones, robots, and self-driving cars launched by tech giants like Amazon. Google and Tesla. But the hype outweighs the day-to-day reality.

AI has a fifty-year history of development, experimentation, and mathematical and computer thinking. It is not an overnight sensation. What makes it exciting is the confluence of large data sets, improved software and platforms, faster and more robust processing capabilities, and a growing group of data scientists eager to exploit a broader range of applications. The mundane, everyday uses of artificial intelligence and machine learning will make a bigger difference to the lives of consumers and brands than flashy apps touted in the press.

So consider this AI reality check:

Big data is messy. We are creating data and connecting large data sets at an extraordinary rate, which is multiplying every year. The growth of mobile media, social media, apps, automated personal assistants, wearable devices, electronic medical records, self-reporting cars and appliances, and the upcoming Internet of Things (IoT) create tremendous opportunities. and challenges. In most cases, there is considerable and time-consuming work to align, normalize, complete, and connect disparate data long before any analysis can begin.

Collecting, storing, filtering, and connecting these bits and bytes to any given person is complicated and intrusive. Compiling the so-called “golden record” requires considerable computing power, a robust platform, fuzzy logic or deep learning to link disparate data, and adequate privacy protections. It also requires considerable modeling skill and a cadre of data scientists capable of seeing the forest rather than the trees.

One on one is still aspirational. The dream of personalized one-on-one communication is on the horizon, but it remains an aspiration. Trigger factors are the need to develop common protocols for identity resolution, privacy protections, an understanding of individual sensitivities and permissions, the identification of tipping points, and a detailed graph of how individual consumers and segments move. through time and space on his journey from necessity. to brand preference.

Using AI, we are in an early phase of test and learn led by companies in the financial services, telecommunications and retail sectors.

People’s Award for Predictive Analytics. Amazon trained us to expect personalized recommendations. We grew up on the idea that “if you liked this, you probably like that.” As a result, we expect our favorite brands to know us and responsibly use the data we share, knowingly or not, to make our lives easier, more comfortable, and better. For consumers, predictive analytics works if the content is personally relevant, useful, and perceived as valuable. Anything less than that is SPAM.

But making realistic and practical predictions based on data is still more of an art than a science. Human beings are creatures of habit with some predictable patterns of interest and behavior. But we are not necessarily rational, frequently inconsistent, quick to change our minds or change our course of action, and generally idiosyncratic. Using deep learning techniques that the algorithm trains itself on, AI can help make sense of this data by monitoring actions over time, aligning behaviors with observable benchmarks, and evaluating anomalies.

Platform proliferation. It seems like every tech company is now in the AI ​​space making all kinds of claims. With over 3,500 Martech offerings plus countless legacy systems in place, it’s no wonder marketers are confused and IT techs stumped. A recent Conductor survey revealed that 38 percent of marketers surveyed used 6-10 Martech solutions and another 20 percent used 10-20 solutions. Cobbled together a cohesive IT landscape in the service of marketing objectives, honing the constraints of legacy systems and existing software licenses while processing massive data sets is not for the faint of heart. In some cases, AI must work around installed technology platforms.

Artificial intelligence is valuable and evolves. It’s not a silver bullet. It requires a combination of trained data scientists and a powerful contemporary platform led by a customer-centric perspective and a test-and-learn mindset. Operated in this way, AI will deliver much more value to consumers than drones or robots.