Artificial intelligence helps to make automated decisions in digital systems that perform tasks that normally require human intelligence. This means that tasks can be carried out efficiently and people can be supported to make good decisions.
Artificial intelligence (AI), is a theme in countless science fiction stories, most often portrayed as part of dystopian societies. In the modern world, artificial intelligence stands behind programs that help us find the shortest path when traveling, translate text, or choose movies on Netflix. Other AI examples range from optimization of industrial processes, via intelligent control of mechanical machinery to robot vision.
Gartner defines artificial intelligence (AI) as applying advanced analysis and logic-based techniques, including machine learning (ML), to interpret events, support and automate decisions, and to take actions." This means that computers and digital systems can learn from their own experiences and appear intelligent.
Machine learning and optimization
High-quality data and skillful data management are crucial to the success of artificial intelligence. Smart sensors and the Internet of Things (IoT) collect data and monitor operation of machines and utilities. With the help of artificial intelligence techniques, such as machine learning and optimization, we create algorithms that provide meaningful insights and optimize actions.
Machine learning algorithms are often divided into three categories:
- Supervised learning uses ‘labelled data’, meaning that each observation has a target value that the algorithm is trained to predict based on input data. For example, an image recognition algorithm can label objects in an input image.
- Unsupervised learning is used to recognize patterns and predict behavior without actively training an algorithm with so-called training data. This is used to group similar individuals or observations, and then predict behavior based on their group affiliation. Unsupervised training can, for example, be used to predict customer behavior.
- Reinforced learning is a machine learning method based on rewarding desired behavior and punishing undesired. A reinforcement learning agent learns through trial and error, by taking actions and improving on them. This is especially useful for problems that can be formulated as a game with clear rules and goals, where the agent optimizes its behavior to arrive to the best solution. Route and maintenance planning are good examples here.
Artificial intelligence and ethics
As a society, we must engage in ethical considerations around the use of artificial intelligence. At the same time, AI is a part of the solution to build a more sustainable society. We should not delay the development of solutions that create value. The use of artificial intelligence in the health sector can, for example, be of great importance for the population's health. And artificial intelligence plays an important role in the fight against climate change.