Artificial Intelligence (AI) : It’s everywhere. The news is full of articles for and against AI, some bad news circulating about how AI is taking over the world, on the other hand how AI can save the world. New products and solutions that use AI are emerging across all industries.
But what is behind this term? What’s up with the news? And where does AI really add value?
In a series of blog posts, we will examine this topic more closely over the coming weeks, explain individual methods in detail and discuss both the added value and point out pitfalls. But first the basics…
What is artificial intelligence?
This question is not easy to answer because there is no uniform definition and understanding is changing dynamically. The definition of intelligence in the human context is still the subject of much debate. If you try to define intelligence for animals or machines, it gets even more complicated.
A dog could not solve the math equation 45 x 223. A person could probably solve them within a few minutes – with the appropriate level of education. A computer or calculator only needs a fraction of a second to do this. So is the machine more intelligent than a human or a dog? Depending on the activity and context, we set different standards for intelligence. This also makes the understanding of AI vague and there are rarely clear answers as to what really is AI and what is not.
AI is generally recognized as a sub-discipline of computer science and deals with the creation of intelligent machines. Intelligent machines are often recognized as such when they can reflect human rationality in information processing, decision-making, and action. It makes sense to divide the concept of AI into so-called “narrow AI” and “broad AI”.
“ Narrow AI” describes an AI system that specializes in performing a specific activity. These include, for example, algorithms that are trained to transcribe speech or to recognize faces in photos. These systems can be very strong in their sub-task, but cannot solve any other, “alien” tasks. The research and developments in this area are already very advanced and there are extremely strong algorithms that can solve individual tasks much more efficiently than humans.
“Breite KI” (in English “Broad AI”), on the other hand, describes an AI system that can emulate almost all human cognitive abilities. This comes very close to the AI machines that we know from films like “Terminator” or “I, Robot”. Machines that behave almost like humans, feel emotions and can solve any task that a human can solve. However, the concept of “broad AI” is still a thing of the future and has only appeared in films so far. Machines that map the entire spectrum of human abilities do not yet exist.
Where does the hype about AI come from?
The term AI was coined as early as the 1950s. However, a lot has happened in technological development since then. One of the basic methods of AI is the so-called “Machine Learning” (ML) , which also includes technologies such as “Deep Learning” (DL) and neural networks. The basic principle of these technologies is learning from data using mathematical and statistical methods.
The reason why AI and ML have become very popular in recent years is mainly due to the technological advances in hardware. This relates to both the performance of processing data and the available storage space for data. For comparison: Today’s smartphone has about a million times the performance and storage capacity of the on-board computer of Apollo 11 (1969).
The availability of “Big Data” and the corresponding hardware with increased performance makes it possible to set up extremely complex ML algorithms and also to process unstructured data such as images, audio and text. This has enabled products and services such as smart assistants, chatbots, facial recognition algorithms and many more to develop.
The figure below shows the development of hard disk storage capacity since 1980. Important: The representation is logarithmic. While the storage capacity was still 0.01 GB in 1980, today hard drives with a capacity of more than 10,000 GB are available.
How can AI be applied?
The use cases of AI are almost limitless. Wherever there is a large amount of data, there is often an AI algorithm. Whether in speech or image processing, the automation of processes, or the targeted use of advertising based on consumer interests: AI opens up new possibilities and opens up further economic potential.
The implementation of AI systems is often easier than you think. There are plenty of providers on the market, from ready-made algorithms to frameworks in which you can train your own algorithms. Of course, there is also the option of having your own algorithms developed by service providers or data scientists. Depending on the application and complexity, different approaches make sense.
The implementation of an AI system usually starts with a potential analysis. Management workshops, for example, are suitable for this. You can focus on the following three things:
- Analysis of AI solutions against the background of the problem
- Analysis of AI solutions against the background of existing data pools
- Examine new AI technologies and find use cases in the company
During implementation, it is important to carefully examine the use case, to understand what data is available, what data flows exist and how they can best be used. Often one comes across situations where the required data is missing and one has to start collecting it first.
Also, the purpose of the different solutions should be well defined, especially in relation to the concept of “narrow AI”. Algorithms are usually specialized for single tasks.
Last but not least, ethical and legal aspects must also be examined. As can be seen from the headlines above, ethical principles should play a crucial role in AI development and should be respected. With a transparent and ethical approach to AI, you can not only do something for your conscience, but also differentiate yourself from the competition.
AI is not new, but it is getting more and more attention due to technological advances. The performance of hardware and AI algorithms will continue to develop rapidly in the future. Don’t miss the connection and find out which AI applications are available on the market and how your company can benefit from them. Liquam will be happy to support you with questions, potential analyzes and implementation. Contact us!
- Vom 13. April 2022