Mantas Lukauskas, KTU Faculty of Mathematics and Natural Sciences (MGMF) doctoral student, Zyro data researcher
Have you ever wondered why your mobile phone automatically corrects the errors while typing? Or why, when shopping online, the e-shop “knows” what else you might want next to the product you are buying?
Both questions have one answer, which is: artificial intelligence! These examples are just a small fraction of what can be done with various artificial intelligence (AI), or, more precisely, machine learning algorithms. While not always obvious, AI affects almost every aspect of our lives. Below are just a few illustrations of this fact.
Everyone who has a relatively new smartphone faces this AI solution many times every day. Many smartphones now have a face recognition feature that helps you unlock your phone without entering any code or tracing the pattern of dots. When the phone detects a face, it scans it and compares it to the owner’s image stored on the phone. The phone is unlocked only when the images match.
AI facial recognition technology can be used in various ways and for different purposes. For example, Canon recently announced that it has introduced smile detection when only smiling employees can enter meeting rooms. Listerine has also developed an innovative smile recognition program: it is for blind people who receive a signal when people smile. This helps them better understand social situations.
Facial recognition can also be used to diagnose diseases that cause changes in appearance. For example, the National Research Institute of the Human Genome uses facial recognition and analysis technologies to detect a rare disease called DiGeorge’s syndrome, with facial recognition helping to diagnose 96% of cases of the disease. As algorithms become more sophisticated, facial recognition technologies will become an invaluable diagnostic tool.
Have you ever wondered why your social network news feed shows particular images, videos, or ads? All social networks, be it Facebook or Instagram, LinkedIn or Twitter, use AI methods to provide personalised information in your news stream.
Based on previous searches, favourite photos, or watched videos, the algorithms suggest what should interest you. AI on social networks can also be used to filter and remove fake news, as well as to protect consumers from cyber harassment.
While we are on the topic of recommendations, we can also discuss the use of AI in e-commerce. Have you noticed that when shopping you are often recommended similar items or the items that you might also need? These e-commerce tricks often are also based on historical data, such as what you bought earlier.
The question is: what if you are shopping for the first time? In this case, there is more than one option for offering you any product. One of the possible strategies is to simply offer you the most popular goods. Alternatively, the algorithm can analyse your metrics, such as age, gender and other data, and then offer you what similar users who have shopped in the same place, bought. Or, maybe, offer similar things to the ones you already have in the cart to make up the right set of items.
There are a lot of options for selling more goods. And almost all of them are based on AI algorithms.
Let’s look further. What do you do if the product or service you bought is not satisfying, and you might just want to ask the seller a question? Businesses receive a high flow of messages and, of course, it is quite difficult to respond to everyone quickly. In such cases, chatbots, which are trained to answer customer questions, are a great help for organisations. Here, the AI serves as a tool for natural language recognition, which analyses the question and links it to the template answers.
Often, the chat robot that submits the answer will check if the customer is satisfied with it. If the customer responds that the answer is not satisfactory, next time the robot will know that and will remember which answers are incorrect. Using robots to answer questions saves staff time and manages customer message flows much more efficiently.
If we are already looking into the answers to the questions that arise, often the first and probably the only way to get information is to search for it on the Google search platform. A search engine cannot scan the entire web and return results without the help of AI, after all, it is not only looking for what you have typed exactly. And what about the ads that seem to “follow” you wherever you go on the Internet? AI is behind all these functionalities.
Every day we send a plethora of emails, write messages or posts on social networks, and while doing this we make spelling mistakes. To avoid them, we have tools that not only check if a certain word is spelt correctly but also whether the sentence structure is grammatically correct. These tools also use AI-based natural language processing models that allow errors to be recognised.
AI solutions can be applied not only to language/text but also to image processing. One of the most common examples of this is social network photo filters. All the tricks, such as putting a funny hat or moustache, or cat ears on your face are also powered by AI, which, depending on the purpose, analyses the image and performs certain functions. That’s how a hat ends up on the head of the person and not anywhere else.
Another interesting example of image analysis could be a heatmap. Its purpose is to use an integrated camera to determine which places on a web page or which photos on that page people will be most viewed. The heatmap makes it possible to understand where to place the most important information or advertisement so as it was most visible.
The above indicated are just a few AI-based functionalities. In reality, we face AI in every field and every sphere of human activity be it medical, banking or financial services. Although AI is still referred to as a future tool, in reality, AI-based solutions can be found in every household today.
