Following the wonderful article by Salvatore Sorrentino on Data Visualization, I would like to once again turn our focus to the theme of using artificial intelligence to support business decisions, not only in those we make as part of our everydays tasks, but also and above all in strategic ones; for the latter our perception of reality often pushes us to use our gut, following our instincts.
I don’t think there is anything wrong with this, on the contrary, I think it is essential to make choices by going beyond what we have as only numerical data, which we know, never lie. But how accurate is our perception of reality? How much have we really been influenced by the facts and not by the noise surrounding the data? Often even our mood can influence our vision of reality and, consequently, our choices. How can we make sure that our decision, even if in contradiction with the data we have, refers to a correct perception of how things really are?
The right data at the right time
In order to be able to make choices based on data, we need to access them in the form and time needed to make the decision. Usually this requires the extraction of the necessary information (which are often located in different sources), their manipulation, and their graphic representation, which highlights the aspects which are relevant to the analysis to be done.
Thinking about this phase, I am reminded of a scene from a brilliant comedy with Robert De Niro and Anne Hathaway, The Intern:
I would say that it is the classic, somewhat fictionalized example of a scene that you will have experienced or seen several times. Do you see something strange in it? If the answer is no, you are reading the right article: this is one of those requests that a good corporate data strategy, combined with the right tools, could have avoided completely. And not only! We also need to ask ourselves how much time passed from when the data was extracted to when it was used. Yes, because there are scenarios where yesterday’s data is fine, but there are also scenarios where it’s too old to use.
In addition, we may need to monitor some information, in order to rectify the choices, or try to predict future values. Can you imagine the sequel to this scene in the film, where Jules keeps asking for variations on the data, or for data related to Ben, going back and forth with intermediate, printed results? It’s okay that this is a comedy, where the situation might be hilarious, but in real life I don’t think it would be that funny.
The importance of data visualization
Another interesting aspect of data analysis for decision making is the visual representation. This is because, as Salvatore has already explained, the representation of the data allows us to highlight aspects that would have been hidden in a large table with many rows and columns. Often we also need to compare different aggregate values, a scenario in which a graphical representation can make a difference.
Wanting to take this aspect to the extreme, I share another scene from one of my favorite films: a “conversation” between Jarvis and Tony Stark taken from Iron Man 3:
I love this scene because, even if it is science fiction, it highlights the two ingredients necessary to extract value from the data we have available: technology and human interaction. It is not artificial intelligence that draws conclusions or makes decisions, but it is the human being, with his ability to see those data with insight and instinct, who takes advantage of the different ways the data are provided.
Beyond the holographic technology imagined in the film, which with a mixed reality device like Hololens is no longer even really science fiction, the idea of being able to immerse oneself in data to allow cognitive processes to find solutions to non-linear problems is something that’s absolutely achievable.
Yes, because what I call the cat algorithm is often valid, which to be honest I did not invent but which I have heard mentioned several times by friends and acquaintances who deal with training. This cute algorithm consists of telling a cat about your problem and being amazed to see the solution in your cat’s bewildered or bored gaze. Obviously it’s just a way to stimulate our mind, by forcing us to say our problems aloud so that, having to explain it, we focus on aspects that we do not evaluate when we’re talking to ourselves inside our heads. For some it works with writing, for others it works by talking to others in the process that is typically called brain storming.
So, having a “cat” to talk to, but which could also show us the data we are thinking about, indicating possible correlations and future values, is a nice stimulation for our mind, which can lead us to evaluate aspects that had been hidden up to that point.
Where is artificial intelligence?
Artificial intelligence is an incredible support in a data analysis scenario for at least three points:
- Natural language interaction
- Data correlation
As we have said several times in this blog, Natural Language Processing and Natural Language Understanding are two obvious applications of artificial intelligence in this field. In fact, they allow us to express requests in natural language and put a Bot in a position to understand our request and respond to us. In complex cases, they can even hold a conversation in order to better understand the request and help us with our work.
Imagine being able to address your Bot with phrases such as “Hi Elly, can you show me last year’s billing information?”, “Can you divide it by quarter?”, “Could you group them by customer?”, “Can you remove customer XXX? “.
It often happens that you have to answer complex questions, where there is no linear correlation between the input information and the solution. In all these cases, extrapolating information related to our requests and proposing it to us can be a great added value in the brain storming phase, which could put us on the right path to solving the problem, by showing us relationships that we had not taken into consideration.
Returning to the previous example, when asked “Hello Elly, can you show me the billing data from the last year?”, The system could also pull out those of the previous or current year. Or you can already propose the same data divided by quarter or by customer.
Unfortunately, technology does not yet allow us to travel in time, but we can use artificial intelligence to make predictions based on the data we have, to help us assess the risks associated with a potential choice.
Imagine, for example, having to assess the impact of the addition of a new employee in the workforce or the loss of an existing resource, and how this could impact the current year’s profit. It is not only a question of assessing the cost, but also the impact on productivity, perhaps taking into account the history of similar operations that had already been made.
Obviously it is no coincidence that I told you about how artificial intelligence can support us in analyzing data to make business decisions. We are actually working on a product that aims to be the key to accessing the company’s information assets, helping us to extract the data necessary to make decisions, in the form and at the times best suited to the different situations we have to face.
We are working tirelessly on the private beta of Elly to collect the first feedback and explore the potential in your domains, if you are interested in participating, subscribe to our newsletter, in order to be notified as soon as registrations are open.
Keep following us!