Today, artificial intelligence is everywhere and all areas and technologies use it for many applications: medical, research, environment, photography, etc. Here is a little overview of the different applications of artificial intelligence by domain.
What is artificial intelligence?
It is “the set of theories and techniques used to create machines capable of simulating intelligence” (Wikipédia).
When we talk about Artificial Intelligence, we usually talk about Machine Learning, then Deep Learning, but also Reinforcement learning.
A. The Machine Learning (or weak AI) which goes through a “supervised learning” by the man: by learning in a very structured way, the AI is then able to infer an object, a situation, etc. This one is very used in the companies.
B. Deep Learning belongs to Machine Learning and works through deep artificial neural networks simulated by a computer process. It is increasingly used, especially in the field of research.
C. Reinforcement learning, the principle of which is that an “agent” explores a space, makes an action and receives a reward according to that environment and its action. He takes note of it and learns by trial and error.
A. “Supervised: all data is tagged and the algorithms learn to predict the result of the input data.
B. Unsupervised: Not all data is tagged and the algorithms learn the inherent structure from the input data. [His result is somehow unknown but his lesser need in terms of human resources gives him an advantage in terms of rapid technological implementation according to his “intelligence”.]
According to a report by Narrative Science, nearly 38% of companies used AI in 2016; in 2018, more than 70% of companies use it.
Among the infinite uses of artificial intelligence according to the domains, it is possible to list:
The medical research: robots with artificial intelligence are able to establish a diagnosis without human help, but also to help elderly or sick, or to detect certain cancers faster than by human intervention only. Are you curious to test Babylon, the “robot doctor”? It’s this way : https://www.babylonhealth.com/ask-babylon-chat
B. An artificial intelligence to count animals of the same species thanks to NVIDIA (source NVIDIA):
Assisting people through voice assistants like Google Home, Amazon Alexa, etc. : Through an ecosystem of “smart” and / or connected products, the user no longer has to carry out the research or manipulation chosen himself, but these assistants do it or have it done in his place.
Securing community platforms: on its YouTube platform, Google has for example set up artificial intelligence tools that can detect violent and / or extremist content and then block it.
Copyright: for several years now, thanks to artificial intelligence, AmazonWebServices or YouTube can automatically detect visual and / or audio content embedded in a video to warn its publisher that it contains works under copyright.
Personalized music streaming: thanks to machine learning made during the musical listening of different users, most music streaming applications are able to offer personalized playlists according to the tastes of their users.
The ecological transition and particularly in agriculture: “[AI] makes it possible to rationalize the farms, to optimize the yield and to contribute to the reduction or suppression of the insecticides and chemical products by detecting the proliferation of diseases or insects as soon as possible. It also helps to better measure and predict ecological disasters in an attempt to prevent them and reduce the risk of damage.” ActuIA.com – https://www.actuia.com/domaine/environnement/
“Using intelligence to make the world a better place” is the slogan of DeepMind, artificial intelligence created by Google. However, artificial intelligence does not always have a philanthropic vocation …
The hidden face of artificial intelligence
The use of artificial intelligence is sometimes disputed for ethical reasons, particularly because of its great potential …
Among these ethical questions, it is possible to mention:
The question of the replacement of man by the machine that lets us ask if it is evolution and adaptation or sometimes simply profitability at the expense of men’s work.
The profiling of personality and privacy to feed the Big Data and offer companies that use it an unprecedented commercial strike force.
The impact of machine errors with AI or else distorted learning for ill-intentioned purposes: MIT students have created a “psychopathic” AI called Norman via a “negative” apprenticeship (from a database of violent and morbid photos only retrieved from Reddit). To discover Norman or to help it modify its idea, it’s here: http://norman-ai.mit.edu/#inkblot
“Smart” decision-making, ie without any human intervention that could be a risk, especially if it is applied to the military field.
Intelligent hacking of personal data for malicious purposes.
The hacking of learning bases for ill-intentioned purposes: for example to confuse the intelligence of autonomous cars and make vehicles dangerous when they are in circulation.