News
Artificial intelligence: Evolution requires imperfection
"Technology has always transformed economic and social structures. Now, with self-powered artificial intelligence, it is generating unstoppable progress that will profoundly change our professional, social and relational habits."
With these words, the President of the Republic, Sergio Mattarella, wished all Italians a happy New Year, also giving a recommendation to those working in the IT sector.
What is AI?
Artificial intelligence is a branch of computer science that deals with creating systems capable of performing tasks that typically require human intelligence, such as machine learning and reasoning.
AI learning occurs through an algorithm that processes data grouped according to specific criteria or through unsupervised machine learning, which autonomously classifies data according to self-defined criteria.
Artificial intelligence transforms big data into new knowledge using algorithms. Artificial neural networks, inspired by our neurons, are used instead of the brain. These systems work differently but effectively.
Technology and science: two different methodological approaches
Unlike science, technology does not rely solely on the processing of data using algorithms. The experimental scientific method involves observing a phenomenon, formulating a hypothesis, carrying out an experiment, analyzing the results, and repeating the experiment.
In many fields, the use of artificial intelligence and support from robots and AI devices enable traditional research fields to expand their horizons, reaching places that are inaccessible to humans.
However, there may be one or more obstacles to machine learning...
Specifically, a machine can only learn if the data it processes is accurate and free from corruption. Corruption refers to data that has been intentionally or unintentionally manipulated by those who create the algorithms and computer sources that underpin artificial intelligence.
Artificial intelligence is based on "acquired" data. However, in a world that aims to evolve and make environmental and social changes for the future, it is important to recognise that AI currently draws its sources from the past and present. The information it processes often reflects old habits, traditions, and stereotypes rather than what we would like to see in the future.
For instance, consider the goal of achieving gender equality among the Sustainable Development Goals by 2030.
If you search for information on the number of women working in motor mechanics by typing "women and motors" into an AI-based search engine, the results returned by the web are often not aligned with the goal of gender equality.
This is due to incorrect classification of the data behind the algorithm, which can lead to... embarrassing outcomes!
The imperfection of human intelligence
Human intelligence not only learns from data processing but also through the senses, emotions, social context, environment, and memory.
Deep learning may enable machines to learn through images, sounds, and sensors, as well as record and accumulate data on human emotions. They may be indestructible and have a very long life cycle, they will be more and more "perfect" machines.
Machines are vastly different from humans, who are full of imperfections but possess a unique intelligence that cannot be easily categorised or replicated by artificial intelligence. The way a person's brain learns is still unknown to us.
Darwin taught us that evolution does not arise from perfection. "It is not the strongest or the most intelligent species that survives, but the one that adapts best to change." Nature has found a way to evolve through the ability to adapt, to change, correct imperfections, and accept others. Without imperfection and the ability to adapt, there is no evolution. Artificial intelligence can correct imperfections, but it always requires the imagination and creativity of human intelligence. Human intelligence is capable of valuing and accepting even what is not perfect and can come up with new challenges.
Download .pdf file
Download file