“I took the liberty to prepare you a bouquet of white tulips for your wife because I know they’re her favorites.” The local shops of yore were popular because of their familiar atmosphere: The clerks and owners would know you, your preferences, and those of your immediate relatives. AI is enabling the same familiar feeling for the globalised digital world. AI allows for mass personalisation.
Companies like Google, Amazon, Cambridge Analytica and several others collect vast amounts of consumer and lifestyle data to build individual profiles. This is then matched to specific online actions, such as push content and tailored ads that are designed to trigger specific emotional responses. The desired outcome of this pipeline is a action, such as buying, liking, forwarding, etc. At Cisco, we do this to estimate the propensity to buy or renew contracts.
How is it possible to generate a pipeline for each individual? This is were AI comes into play: Techniques such as Deep Learning excel at using very small differences over large quantities of data to build mathematical models, such as personal profiles. Each profile can easily be based on 5,000 input parameters, which is to be multiplied by the number of users, ideally in the order of the millions. You can see that just the creation of individual profiles is a matter of billions of data points. Too much for human engineering, but not for AI.
As result, AI platforms for sales and marketing claim to deliver revenue increases more than 30%, sales increases more than 20%, and profit margin increases almost 40%. The magnitude and effectiveness of the phenomenon shows that we are witnessing a market transition in sales and marketing, with AI becoming the Dale Carnegies or Don Drapers of the modern age. Sales are becoming more and more personal and accurate, but with a more than 1000-fold higher throughput than local shops of door-to-door salespeople. The same for marketing campaigns, with the additional benefit that AI systems are becoming increasingly capable of creating videos, text, audio, and images that elicit specific emotional responses.
Besides the benefits, what are the drawbacks of such large-scale social engineering? For one, the expectations of customers will rapidly increase. It won’t be sufficient to answer within 24h or rely on personal relationship (exceptions of course apply). Exactly like it was suddenly an imperative to have a website to be taken seriously. This is not actually a drawback of the market transition, unless companies and their people fail to embrace and adapt quickly enough.