What is artificial intelligence and how does it work

by
What is artificial intelligence and how does it work

Until you got to this story, you probably went through some processes that involve artificial intelligence (AI). If you received this text via Twitter or Facebook, your interests and your browsing profile may have been what got you here.

Have you also noticed that Netflix movie and series covers change over time? Understand how it works and how artificial intelligence is applied in everyday life.

What is the difference between bot and algorithm?

Bot is a robot or machine programmed to do something. It can operate autonomously or semi-autonomously. On Twitter, for example, a good example is an automatic profile that maps the agenda of a minister and publishes it on the network.

In the case of chatbots, these are robots with an extra layer of intelligence for interacting with the physical world in some way. As an example, we have the virtual service of an e-commerce. In some cases, the robot identifies words to give the closest answer to the consumer.

The algorithm can be explained as a sequence of actions to achieve a goal. If a music stream, for example, is in new song suggestion mode, the algorithm will identify a sequence that has to do with the user’s taste.

What is artificial intelligence?

AI (artificial intelligence) is a way of imitating or reproducing human intelligence and our learning process, and may even be able to make decisions. It is based on algorithms to develop systems to complete tasks, automate processes or learn and reproduce logical reasoning, and has a more developed “complexity of thought”.

What are the types of artificial intelligence?

Artificial intelligence can be strong or weak.

THE strong artificial intelligence has the ability to understand or learn any skill, just like a human being. She doesn’t exist yet, and is only represented in science fiction movies. It would be something like the assistant computer HAL 9000, from the movie “2001: A Space Odyssey”.

THE artificial intelligence weak is currently used. She is able to learn skills for a specific use.

How is artificial intelligence created?

Artificial intelligence works through an algorithm. There are different methods of teaching an algorithm. Everyone, in some way, has humans in the process, with greater or lesser involvement. The main learning approaches are: supervised, unsupervised and by reinforcement.

  • supervised learning

In this type, data that is labeled is presented to the algorithm, that is, data that has undergone some type of supervision.

Example: Imagine a computer learning to recognize a yard with various breeds of dogs. In supervised learning, a field with their description is inserted.

  • unsupervised learning

In unsupervised learning, the algorithm does not receive labeled or categorized data. It ends up finding patterns in the available data and makes decisions based on the patterns found.

Example: Still on the recognition of dog breeds. The algorithm will have no idea of ​​the dog concept, so it won’t be able to label it, but it will be able to group them according to color, size, fur size and other differences.

In reinforcement learning there is no demand for large amounts of data for its training. As the word goes, the algorithm learns by trial and error, until it understands what it should do.

Example: An algorithm learning to play video games. Using the pixels on a screen, a character can learn from their movements whether they are performing well or poorly. Through this information, the computer learns which moves are best to beat the opponent.

How is artificial intelligence used in everyday life?

Transport applications have large volumes of information. Waze learns routes and traffic intensity based on information from active users, and uses this data to direct faster routes according to user-determined functions.

Waze collects data from every road a user drives with the app open.

The next time a user takes a specific route with the app open, the app will know how to compare the data between each possible route and will be able to suggest the ideal route.

Robot vacuums have gained popularity in recent years, but they don’t have a camera to guide them. The machines have sensors that are able to scan the space and, depending on the model, determine its trajectory according to the level of dirt in each environment.

Robots can detect hazard, like a ladder. In addition, the device’s sensor measures the distance between the robot’s base and the ground, making it turn around and not cause an accident.

Let’s take as an example the documentary series “Andy Warhol’s Diaries”, which was released by Netflix this year. To reproduce Andy Warhol’s voice, it was necessary to use a number of pre-existing audio from the artist, who used to record his thoughts between 1970 and 1980.

Through deep learning, scientists were able to gather a set of information such as voice characteristics (intonation, frequency and timbre) and phonemes. With this dataset, it was possible to formulate a voice with the same traits as Warhol.

Netflix’s algorithm often uses user consumption data to generate new covers.

The platform uses a system based on Aesthetic Visual Analysis, a type of tool that scans the entire video and searches using a series of algorithms, choosing the best images from a series or movie according to the user’s taste.

Example: if you like actress Drew Barrymore, even if she appears a little in a movie, you can find her on the cover.

Sources: Diogo Ortiz, professor at PUC-SP, IBM, Waze, Google, Vox

You May Also Like

Recommended for you

Immediate Peak