We are often led to consider the artificial intelligence as a sphere far from the daily application. The situation is different and there are realities, more or less established, that make extensive use of it. Here are some examples with which we are confronted every day and others more intriguing with which we will be confronted in the near future.
We are still far from being able to consider artificial intelligence as a sentient intelligence comparable to the human one, capable of thinking autonomously with a critical spirit and according to conscience.1. While there is no shortage of controversial cases2, artificial intelligence today is a set of precise learning algorithms or simulators (developed by man) that are able, thanks to the computational power of modern computers, to perform defined tasks in defined areas.
contemporary AI is based on Machine Learning o Deep LearningIt is a set of methods thanks to which the machine (through algorithms or the decisional simulation of the neural networks) learns (analysing massively the data that are fed to the system) and trains itself (thanks to prescribed input rules), correcting the errors in order to succeed in time, after a great amount of analysed data and on a statistical basis, to carry out a certain activity in autonomy.
Some usual applications
- Predictive research
- Virtual assistants (Siri, Alexa, Google)
- Recommendations (products - music - movies - social media)
- Online advertising
Some unusual applications
- Artistic experiments
- Artificial intelligence in gaming
- Search scientific
How does it work?
The general operating scheme is as follows:
- An AI-based technology infrastructure capable of receiving input and exporting output is created
- Inputs are the data that are given to the system, big data, are a concrete example of this
- These data are analyzed and processed by the algorithm (machine learning) or by the neural network (deep learning) that elaborates them.
- Depending on the predefined goal, the reprocessing generates an output. This clearly varies according to the previously designed infrastructure.
To give a few examples:
- In the case of Google's predictive searches, the suggestions proposed (output) are the most typed words in that particular search, with that particular mix of words by the users (input)
- In the case of Netflix recommended movies, the recommended movies (output) are derived from the taste of the viewer (input), based on a number of factors such as movie genre, actors, stylistic preferences etc.
- In the case of the AI that beats the human video gamers, through the analysis of the recursive schemes of game by the video gamers (input) the machine decides autonomously the best strategy (output) to beat the adversary
- In the case of the analysis of the lunar craters, the AI is instructed to understand what a crater is (dimensions, colour, dissimilarity etc) analyzing the photographic material (input). Subsequently, it is able to extrapolate, from other photographic material, what it interprets as a crater (output).
In this technological context the projects of artificial intelligencemore or less complex, based on machine learning and deep learning techniques.
The applications are of various kinds, some business ready: such as tech startups that make use of it, to solve problems in a given industry, thanks to the vast amount of data available (Big Data) or being generated (IoT, Web etc).
More more visionary as in the examples proposed in the field of health, art or astronomy which, by exploiting the same technologies, open the door on the one hand to the acceleration of scientific progress and optimization of processes, on the other hand to a whole disquisition of social and philosophical questions.