AI is everything from using the raw power of computers to automating simple things, to superhuman skills. Large data volumes are often included in the picture. Here are some hot applications for AI.
What makes the development and interest in AI (artificial intelligence) accelerate at an ever-increasing rate right now? One answer to that question is that it is because there are now plenty of examples of useful AI solutions and even more ideas about solutions that feel reasonable to create.
Why is this happening right now? Here are five reasons why:
- There is now great access to large amounts of data, which facilitates the creation of many types of AI solutions.
- Relatively cheap cloud services make it possible to create AI solutions at a reasonable cost.
- A wide range of frameworks and tools simplifies the work of building solutions.
- Really fast processors adapted to AI make it possible to run completely new solutions.
- The interest is self-generating. Much interest gives birth to even more interest.
How did we get to this point? What does the AI look like?
If we ignore the thought of building reasoning machines as early as the 1300s, it can be said that AI (artificial intelligence) gained momentum in 1943. Then the American scientists Warren Sturgis McCulloch and Walter Pitts presented a design for “artificial neurons”, ie an artificial intelligence equivalent to one of the most important constituents of the human brain.
The research area AI gained momentum in 1956. You can read on Wikipedia, at a conference at the University of Dartmouth College in the United States. Since then, the development in the AI area has looked like a roller coaster.
In the mid sixties, AI labs were established all over the world. Then it was quiet for a while, to regain momentum in the early eighties when rule-based expert systems became popular. That interest gradually faded away and today expert systems are hardly seen as AI solutions.
Around the turn of the century, AI solutions for logistics, data mining, medical diagnostics and other areas began to be used. This development has led us to date, with an interest in AI approaching the boiling point.
What is AI?
But what is AI? There are certainly almost as many definitions of AI as there are experts. We can start by looking at some areas that can definitely be seen as artificial intelligence, but which do not always appear in the categorizations of modern, computer-based AI:
Artificial general intelligence is an area where machines are able to handle everything that humans can handle. The type of AI solutions that are attracting interest from business today is more about building solutions to solve specific problems.
Artificial general intelligence is sometimes called “strong AI” or “full AI”. The more problem-oriented variant is called “weak AI”.
Here are several partially contradictory definitions.
IBM is the company that has highlighted the concept the most, and then it is about building solutions that can handle the same things as a human being, but partly in different and better ways.
Robots. AI solutions of different types are often important components for building robots, but the intention of using them is to create a whole that mimics people on more than just the reasoning.
The more problem-oriented “weak” AI solutions are often found in one of these areas:
Massive automation. An example is sorting completed forms on a website, based on the significance of what is written in them.
Predictions that you do not really understand how they go about. Everything from stock levels in stock and smart route planning, to expected stock prices.
Tasks that include such a large amount of data that it is obvious that a human cannot perform them. For example, discovering which features of an online game are popular, in real-time.
Pure compensation for a human, such as a self-driving car checking the surrounding traffic, instead of a human driver using his eyes to do so.
There is no truly accepted categorization of AI solutions, but many are often partially overlapping. A fundamental problem is that technology solutions and application areas are mixed together wildly.
But it may not matter much overall. The most important thing to know right now is that new AI solutions make entirely new applications possible. This is not always due to the use of new technical solutions. Sometimes it is about cheaper hardware or new cloud solutions making it reasonable to actually use old solutions.
Most of all, it’s probably because so many are working so hard to build AI solutions today. This work effort has not been discontinued before.
Anyway, here’s an attempt to list interesting areas of application for AI right now:
Reasonable applications. Chess computers are a simple example. Here you can sort data analysis solutions (data science), for example for forecasting and different types of probability-based solutions. Perhaps planning can be highlighted as a collective concept. An example is to predict how many goods of a certain type need to be purchased at a department store. The solutions are often based on historical data or real-time data streams, but not always.
Natural language, with connection to databases. Typical examples are chatbots on customer service sites and care sites.
Perception. Using cameras, microphones, motion sensors, sonar and similar equipment to get an idea of the physical environment. Image recognition and interpretation are used, for example, in self-driving cars. Another example is identifying threatening people in public groups.
Robots. Is very much about making decisions based on perception, to decide how robots should move and manipulate the outside world. A robot can be anything from a human-like android to a nurse replacement, to a “box” that performs a repetitive step in a factory.
There are several more examples of application areas for AI and they are often combined in ready-made solutions. But the list above covers a large number of interesting opportunities.