Abstract from the book by Joseph Sassoon // Researcher and author expert in brand and storytelling

Getting a car to drive itself actually requires a lot of intelligence. If you want to prevent it from running someone over or crashing into a wall, you need to fill it with algorithms and an incredible amount of data and information.

But today’s artificial intelligence is not limited to areas of special application like this. Although we are often unaware of it, it accompanies us in a rapidly growing number of everyday situations. When Netflix presents us with lists of films that we might like, this is no accident: it is the consequence of the use of AI, i.e.: of sophisticated software that sifts through our previous choices, compares them with those of other users with similar tastes to ours, and proposes films that they have already seen and we have not.

The fact that AI is used by a platform like Netflix is, in itself, an interesting case of the intersection of technology and storytelling.

In a context in which artificial intelligence offers automated solutions to a multitude of human needs, the key question is this: how much can and will AI affect the specific content of storytelling, the ways in which stories are conceived and told?

As I started investigating this subject, I soon came to realise that it is surrounded by a lively debate among experts and storytellers, but is mainly discussed in articles or blogs online and, to my knowledge, there is still no comprehensive analysis in book form.

Although it may seem easier to tell a story than to drive a car, in actual fact getting a robot to create a story poses a number of very particular problems.

Artificial systems struggle to cope with the huge wealth of language, the variety of contexts and the complexity of emotions.

I can say that these problems mainly concern the difficulty experienced by artificial systems, which usually have no experience of the world, when it comes to absorbing the huge wealth of language, the variety of reference contexts and the complexity of human emotions. While this is true, it is also true that many people are working on solving these problems, and that major world players are investing significant sums of money in this area.

First of all, what exactly is artificial intelligence? It can be defined as the ability of computer systems to perform tasks that normally require human intelligence.

In order to function, AI then uses algorithms. What is an algorithm? While the word may sound a bit mysterious, its meaning is quite simple: an algorithm is a sequence of instructions that tells a computer what to do. 

Although separate from and independent of humans in many respects, artificial intelligence obviously originates from human intervention.

Another important expression often used is machine learning. This refers to a set of algorithms that analyse data, learn from it and then apply what they have learned to make informed decisions.

An easy example is that mentioned before of Netflix or an on-demand music streaming service: in the latter case too, the system makes decisions regarding which new songs to suggest to you based on how machine learning analyses your preferences and those of other users with similar preferences to yours.

There is also a subset of machine learning called deep learning. It differs from the broader definition in that it is characterised by a greater degree of autonomy. As explained in a Zendesk article (2017), generic machine learning systems become progressively better at what they do, but they always need some support: if an algorithm results in an inaccurate prediction, an engineer has to step in to make adjustments.

But a deep learning model, which works on the basis of a layered structure of algorithms called a “neural network”, can manage on its own: its algorithms can independently determine whether or not a prediction is accurate.

With deep learning, therefore, the learning process escapes human control and is of higher quality (nevertheless, deep learning is still a form of machine learning).

Robots are now learning what stories are, how they are built, what makes them attractive or effective, how they can be charged with emotion

We are at a stage in which robots are learning what stories are, how they are built, what makes them attractive or effective and how they can be charged with emotion, and they need machine learning for this.

The progress of machines in the field of storytelling will also tend to have a massive impact on companies and brands. Those who understand this first have a good chance of gaining major competitive advantages.

Source: Sassoon J., “Storytelling e intelligenza artificiale“, Franco Angeli editore, 2018