From Artificial Intelligence to Augmented Intelligence

Train Artificial Intelligence to provide the best customer experience

Using AI to devise the experience design


At the turn of the Customer Age, modern Marketing stands out because of its peculiarity of being increasingly data-driven and its desire to gather data from consumers to create one-to-one communication strategies that can adapt to the specific needs of different targets. And it's thanks to technology - in this case Artificial Intelligence (AI) – that we can make sense of the huge amounts of data (big data) handled every day.

Giving new meaning to the data, which is analyzed and processed by the AI, helps us - as Umberto Basso, Managing Director of AKQA, notes - devise the experience design, that is the experience of the user during their customer journey, through the contact points with which the Brand communicates.

There are two aspects to keep in mind. First of all, this interaction must be centered on the needs expressed by consumers (that's why a customer-centric vision is required to ensure you are always aligned with your audience!). Second of all, it must always be consistent, i.e. it must appear uniform and seamless among the crowd of devices and touchpoints connected to the advent of the digital revolution.

AI algorithms work tirelessly to extract data and succeed, with insights about the purchasing behavior of customers collected in real time, in constantly improving the experience by sending personalized messages (we know that customization increases engagement!). Data is the fuel - some call it the "new oil" – which powers AI. The more data we have, the more Artificial Intelligence manages, through an automatic process, to learn and adapt to the user in front of it.


When AI ends up looking bad


As good as AI is at analyzing huge reams of data and even if it knows how to search for patterns and correlations invisible to the human eye (or obtainable only after several efforts that take time and energy), without constant training it risks making terrible blunders. During the event "Value creation strategies and Content Intelligence: measuring is possible", Umberto Basso of AKQA cited the example of AI failing to identify the photo of a person of Chinese nationality because it categorizes them as "having their eyes closed".

And it doesn't end here, the list of public and embarrassing examples of AI gone bad is quite long. The limits of the algorithms have been seen in the Photos app of Google, where black people were tagged as "gorillas" and even today the problem has not been solved, given that no result is shown when you search for the term “gorilla” on the Google Photos search engine. The risks of altering the user's perception of the Brand ("gorilla" is a very serious insult!) make it clear that the stakes are very high.

Furthermore, if we consider how AI engines are now incorporated into a wide variety of organizations and how they influence decision-making processes, it makes us realize how much they can influence our own culture. An example is the case of Facebook's censorship of the famous Pulitzer Prize photo portraying a naked child fleeing from napalm bombs during the Vietnam War.

Some say that the future of AI and its benefits are directly linked to data quality. If AI engines learn from the data, it means that the data is not enough or not accurate. Consequently, if the data is of questionable quality, so will be the results and the analyses produced by the AI. It therefore becomes a priority to concentrate on the reliability of the data with which the AI is educated.

In a scenario where there is more and more talk of "algorithmic accountability", i.e. holding the developer of the algorithm responsible for any problems that occur, Content Intelligence, that is Artificial Intelligence applied to content, offers the undeniable advantage of collecting First Party Data on the interests of consumers based on their use of the content. This is information about the audience that is extrapolated directly and in real time from the touchpoints with which the Brand communicates and which is directly owned by the company.

This ensures that the AI engines work on valuable data that is unambiguously correlated to the users and constantly updated: this way marketing automation initiatives that are calibrated on the single customer view (a snapshot of the entire visualization path of the user, in which the data collected by the CI is enriched by CRM and other data management systems) provide a personalized customer experience that really meets the expectations of our consumers.


The next evolution: Augmented Intelligence


Umberto Basso of AKQA has stressed the importance of the concept of Augmented Intelligence, a process in which AI is seen as a tool capable of enhancing and increasing human skills while still being under the control of the company that integrates it into its systems. The marketing teams, in fact, must ensure that they "train" the algorithms with a clear vision of the final objective, that is, the offer of the best possible customer experience.

In this context, taxonomy plays a very important role: it's the tag dictionary on which AI relies to organize, analyze and automatically classify each one of the Brand's digital assets. In this case there must be previous work to define the taxonomy, in order to understand, if you want the AI to automatically perform these operations in the best way possible, which tags are most appropriate for the products or services related to your company. This is precisely one of the tasks of the Content Intelligence Manager, the cross-departmental figure who is responsible for improving the content management processes in order to ensure the benefits of CI.

With THRON, the Intelligent Dam (with integrated CI), the official taxonomy on which the AI engines will feed can be established by the Sales and Marketing offices themselves, so that the tagging will respect the company strategy 100%. And the advantage linked to this functionality (called "Tag center") entails the option of inserting the various tags by hand with the possibility of modifying or merging them (they are standardized when they refer to the same content) because, once modified, the change will in any case propagate automatically on all related content without losing the historical data accumulated up until that moment.

At this point, with the certainty that the AI is organizing content in the best possible way, it will be easier for marketers to concentrate on analyzing the data related to the use of that content and on optimizing the editorial strategies of their Brand in order to increase engagement and conversions.

Do you want to learn how to create the right taxonomy for your Brand?

Download white paper