“Artificial intelligence” (AI) has become a buzzword in many technology-related communications these days. Autonomous cars, autonomous buses, autonomous planes, drones, robotics, traffic management, smart city, predictive maintenance, etc. are few things that are alluded to when talking about AI. It seems futuristic, although it may be surprising to find that AI is already part of the past and present, and that we are surrounded by it in our daily lives.
Do we really know what Artificial Intelligence is? Do we know the difference between symbolic artificial intelligence and connectionist artificial intelligence? Can we develop smart solutions without using Machine Learning and Deep Learning? What are the sectors and areas of application of AI?
Artificial Intelligence (AI) can be defined as the simulation by computer systems of human intelligence processes such as learning and reasoning. Since its beginnings, which date back to the article “Computing Machinery and Intelligence,” published in 1950 by Alan Turing, AI has had its ups and downs associated with advances in technology.
Between the 1950’s and 1980’s, symbolic AI was the dominant paradigm in AI research. Symbolic AI is the term for the set of all research methods in artificial intelligence that implements symbolic reasoning methods called rule engines, expert systems or knowledge graphs. It represents the more traditional approach of coding a model of the problem and expecting the system to process the input data according to that model in order to provide a solution.
Between 1960 and 1970, researchers were convinced that symbolic approaches would ultimately succeed in creating a general artificial intelligence machine. However, in the late 1970’s, researchers decided to abandon the symbolic approach, in large part because of the technical limitations of this approach. It was replaced by an AI which exploits the statistical and mathematical focuses on specific problems, rather than the general artificial intelligence.
In the 1980’s, machine learning was growing, especially with the renaissance of connectionist AI via the machine learning. The machines started to deduct “rules to follow” only by analysing the data.
Machine Learning is a set of learning algorithms divided into five categories depending on the type of learning used: supervised, unsupervised, semi-supervised, reinforcement and transfer.
Machine learning systems must be powered by structured and categorized data that allows it to learn how to classify similar new data. To improve performance, the intervention of a human expert is possible to indicate to the system the correct categories of incorrect classifications.
Deep Learning emerged in the early 2010’s. It is a subcategory of Machine Learning as it relies on unsupervised learning. Deep Learning is particularly suitable for complex tasks, when all aspects of the object to be treated cannot be classified upstream. Deep learning systems do not require the use of structured data. They are able to determine the discriminating characteristics themselves. In each layer of the neural network, the system looks for a new object-specific criterion as the basis for deciding which classification to keep for the object at the end of the process. Using Deep Learning, the system itself can identify the discriminating characteristics of the data without the need for human experts to perform a prior classification.
In recent years, the development of these different methods of machine learning has evolved thanks to a favourable environment combined with the continuous fall in the price of computing power and the spectacular increase in the flow of data, mainly from social networks. Therefore, the relevance of using AI technologies is quite reasonable, as it involves a large amount of heterogeneous data, including structured and unstructured analysis, in near real time.
Advances in artificial intelligence technologies are beginning to provide humans with the tools to facilitate decision-making, as well as the performance of certain tasks in several fields. Here are a few examples from our daily life, as well as in the industrial setting where artificial intelligence finds its place.
Online shopping and advertising
Artificial intelligence is often used to provide personalized recommendations to users based on their search history, purchase history, or online behavior. AI is very important in the world of commerce.
Depending on the personal data of the people visiting the e-commerce site, their needs and expectations will be different. Is it a woman or a man? Parents? People used to buying sportswear? A person who prefers comfortable clothes or luxury clothes?
When looking at an advertisement on the Internet, how many times have you said to yourself, “I thought I would buy this product”? It’s artificial intelligence at work.
There is no magic here because AI cannot read your mind. It is simply a system that can track your online activities and then exploit them. Products that you browse on different shopping sites or search engines are tracked, and advertisements related to those products are delivered to you.
Data science techniques and behaviour analysis are there to offer technological bricks to existing tools in companies to implement this kind of customization.
Smartphone & Social Networks
It seems like a challenge to imagine our lives without cell phones. The various applications on our mobile phones are now an integral part of our daily lives. Among these applications, some are based on artificial intelligence.
The smart assistants built into our phones, such as Siri, Alexa, and Google Assistant, are the most obvious examples of artificial intelligence that most of us know and use.
More and more mobile technology platforms are developing solutions that use artificial intelligence to manage different aspects of devices, such as battery management, event recommendations, etc.
We go from smartphone apps to artificial intelligence in various social media apps. Whether it’s your Facebook, Twitter, Instagram or other influenced platforms, AI controls the flow you see while browsing those platforms or the notifications you receive.
It takes into account your tastes and preferences, your history, etc. Organize the information in a way that you find more relevant and that you tend to consult.
Personal digital assistants, voice control & automatic entry
Smartphones exploit artificial intelligence to provide products that are as relevant and appropriate as possible. Virtual assistants answer questions, provide suggestions and help manage day-to-day tasks.
Do you want to know how AI is used for voice control?
No need to look far. Just look at your own mobile device. The integration of AI makes typing more comfortable. It can predict words, phrases and emojis based on your common usage and writing style. This is beyond the predictive writing used in previous cell phones, and it is recommended to be more context-sensitive and to imitate your writing style.
When we think of the use of AI in smart home development, naturally we think of Alexa and Bixby.
However, these artificial intelligence applications are not limited to these intelligent voice assistants.They use AI to automatically adjust the temperature of a device at constant temperature to save power by automatically turning it on/off lights according to the presence of people, smart speakers, applications that change the color of the light according to the time of day, etc. These are apps which use artificial intelligence to make the home smarter.
Artificial intelligence is constantly evolving, and more and more solutions are being developed to understand our behaviour in order to act accordingly.
The idea of artificial intelligence brought on the concept of its use for larger-scale surveillance. If ethics is a controversial subject, it is well known that artificial intelligence is used more and more in this field. Monitoring the huge stream transmitted by various cameras and other devices is not only a tedious activity, but also has its limits. Artificial intelligence uses technologies such as facial recognition, object recognition and location recognition. It can monitor and analyse incoming entries.
Banking is one of the fields that adopted technological inventions earlier than most others.
Banks use artificial intelligence in many areas, including to detect fraudulent activity, analyse customer investment trends, provide customer service, and more.
When you use a new device to make a transaction, have you received a notification from your bank? This is a use case of AI to detect any potential fraud. Notifications received from banks and financial institutions regarding their services and products are examples of AI understanding your preferences, requirements and financial strength to recommend related products.
Thanks to advances in Machine Learning, Deep Learning and Big Data, artificial intelligence continues to gradually change the medical world.
Artificial intelligence enables more precise and detailed medical diagnosis. AI is also enabling the development of surgery, radiology, therapeutics and the management of hospital traffic in hospitals.
Advances in artificial intelligence will make the medical diagnosis of specific diseases more precise and detailed. Thanks to big data, large amounts of data can now be collected in a simpler, faster and more efficient way. Obviously, all of this is done with respect for patient privacy.
For example, in order to combat the spread of the coronavirus, researchers experienced in data analysis and data science have developed new methods to detect the virus by exploiting only the sounds of coughing. Artificial intelligence is also used to study vaccines and treatments for the disease. Not to mention that artificial intelligence makes it possible to fight against Covid-19, by using it in the production of thermal imagery and in other situations in airports.
In the medical field, AI can detect infections by performing a CT scan of the patient’s lungs. AI also made it easier to collect data to track the course of the infection.
The progress of artificial intelligence and Big Data goes well beyond the advancement of medical diagnostics. Indeed, in the near future, the branches of surgery, radiology and therapy will be subverted. The main objective is to improve the accuracy of treatment and prevention of various diseases that affect the world population.
Researchers are studying how to use artificial intelligence to analyse large amounts of health-related data to uncover recurring patterns, generating new findings and ways to improve personal diagnosis.
Here is an illustration: researchers have developed an artificial intelligence program capable of responding to emergency calls. The program is expected to detect cardiac arrest cases on calls faster and more frequently than medical dispatchers. Another example is the KConnect project, co-financed by the EU, which develops multilingual search and text services to help people find the medical information that best suits their needs.
Smart cars are another area where artificial intelligence is increasingly popular in our daily lives. Not just companies like Tesla are at the forefront of automotive automation applications. Many car manufacturers are also planning to incorporate artificial intelligence into cars in order to provide services to drivers.
Information is shared and communicated between cars to drive better in traffic. Real-time updates of traffic entrances and roadblocks are transmitted immediately to remind other vehicles on the network to authorize rerouting.
Although autonomous cars have not yet become the norm, our cars have already used AI-based safety features without necessarily using machine learning or deep learning. For example, Automatic Emergency Braking (AEB) is a device that allows the car or truck to brake automatically when it detects an imminent collision with a detected vehicle, pedestrian or other obstacle.
The AEB system is designed to handle different road scenarios. First, it will warn the driver of obstacles in front of the car. If the driver does not take action in time to avoid the collision, the AEB system will automatically apply the brakes with varying degrees of force according to the intelligent speed algorithm.
Other smart options are more and more present in cars like automatic parking, voice control, gesture control and drowsiness detection.
In the aviation industry, AI can be used to provide automatic communication with control towers, automatic take-off and landing of aircraft, automatic routing of aircraft on land, and inspections for fault detection (predictive maintenance).
In the aviation sector, the development of artificial intelligence enables airlines (1) to increase their revenue by quickly understanding customer preferences, optimizing prices in real time and determining the preferred destinations of specific audiences; (2) to optimize the use of airspace through predictive maintenance; (3) to track baggage in real time, accurately estimate the amount of fuel needed for the flight and reduce costs; (4) to ensure customer satisfaction by analysing customer feelings or analysing trips; (5) to implement risk management models and strategies by integrating the fatigue estimation model into the crew planning software. Therefore, the schedule can be adjusted based on the estimated fatigue risk of each pilot.
Day after day, we discover that smart solutions help people become more productive. It offers them services to access information more easily and make decisions quickly.