What is Artificial Intelligence?
AI, to put it simply, is the attempt to transfer human learning and thinking to the computer and thus give it intelligence. Instead of being programmed for every purpose, AI can find answers independently and solve problems independently.
The aim of AI research has always been to understand the function of our brain and our mind on the one hand and to be able to artificially recreate it on the other. The dream of artificial intelligence is older than the computer itself - be it “Frankenstein's monsters” or artificially created people like the Homunculus.
Artificial intelligence in science fiction and reality
When we talk about AI in today's world, however, it has little to do with what we know from films and books. In real life, we only encounter AIs in secret - when new products are recommended to us on Amazon, when people are automatically recognized in photos or when we chat with “Alexa” or “Siri” on our cell phone.
Define the term AI
So what is an AI? It's hard to explain clearly. It can be said that there is no generally applicable definition of artificial intelligence - because the term intelligence is not clearly defined either.
That's why we're trying to approach the term differently: In German, a distinction is often made between strong AI and weak AI when it comes to the definition of AI. To put it simply: Strong AI means what we know from science fiction. A machine that can solve problems of a general nature - that is, every question you ask it. So far it is still pure fantasy and will remain so for decades or centuries.
On the other hand, we deal with weak AI in everyday life: these are algorithms - and nothing else is an AI, a very complex algorithm - that can answer specific questions whose solutions it has learned independently beforehand. She has no consciousness of her own and shows no understanding. (Well, she might share the latter with some strong AI like the Terminator).
Strong AI belongs in the field of science fiction, weak AI is part of our everyday life
What is AI?
In the following, we will only talk about weak AI because it is ultimately the only commercially relevant form today - we find weak AI in our mobile phones and computers in everyday life.
What is the difference between an AI and a simple program? Usually, a programmer writes code in a language of her choice, which consists of a set of instructions of any complexity:
If this, then that( programming).
if (condition) {
// block of code to be executed if the condition is true
}
When the user presses "Send", send the e-mail to server X.
Such a system is also called rule-based. With artificial intelligence, the programmer does not specify every single step but instead writes an algorithm that can create these steps independently. Why is that important? Because certain problems are so complex that it is impossible to code for them.
An example of this is image recognition, which is used in social media such as Facebook: no programmer in the world can write a set of instructions that always recognizes how I look, regardless of whether the photo was taken at night, on the beach or in the car - In a rule-based system that would be completely impossible, because the programmer would have to know all the cases in advance and laboriously type them in.
An AI does not know every picture of me either, but it can learn from several existing pictures of how I look and then transfer this rule to new pictures and recognize me. And not just with me, but with billions of faces in fractions of a second. An AI is therefore able to deal with previously unknown data, to find patterns, or to derive actions from them. It learns independently from the data available to it - what it learns, however, is determined in advance by humans by designing the AI. This makes them far more powerful than rule-based systems, as they can - within a certain framework - react to previously unknown situations and learn from experience.
What can an AI do?
The possible uses of such AI systems are gigantic and most people are not even aware of it. It will revolutionize our economy. AI can extract information from data that a human could never grasp, for example, because it is too numerous or the underlying patterns are too complex.
Imagine if YouTube employees had to manually view every uploaded video and check whether it contained prohibited or stolen content. 400 hours of material are loaded onto the platform every minute. The group would need 72,000 employees alone who watch videos non-stop for 8 hours a day in order to keep up with the viewing! An AI manages this during the upload process, virtually in real-time.
Artificial intelligence like these is very good at capturing unstructured data as well. This could be, for example, images, videos, texts, or sound recordings - data that cannot simply be searched by computers because they do not have a uniform form or are not measured values from sensors. A conventional search algorithm (such as when you enter CTRL + F on this website) can find the title of a picture (a structured date), but not whether Susie Mustermann is shown in the picture - this information is nowhere, it is part of the Image content. An AI can do that.
Of course, AI is also used to sort structured data and search for patterns. The current upswing in the field of AI takes advantage of the fact that unstructured data is generated much more frequently: it makes up around 80 percent of all data and has only been available in large quantities for a few years- with the boom of the Internet, Industry 4.0 and the massive availability of (cloud) storage. Many companies do not even know what data they have and what potential for added value they have. Be it machine data, audio recordings of customer calls or recordings of transport routes. You can read a few examples later. Only the massive availability of data in connection with the massive progress in computing speed has led to the fact that AIs have become usable on a large scale in recent years.
What can she not do?
AI is not a general problem solver - not yet. It can process data extremely well and recognize patterns, but it cannot understand it. Artificial intelligence has no "common sense" - no understanding. If it comes to wrong conclusions due to insufficient data or poor programming, it does not recognize this (see section “Artificial intelligence and humans”). It can only provide answers to the specific questions for which it was programmed.
Ideal for an AI: Automatic quality inspection on the assembly line in a factory using optical sensors© WFB Jann Raveling
Examples of AI projects
AI has long since found its way into our everyday lives. The example of facial recognition on social networks is one of many. Another is voice assistants on our mobile phones - Siri, Alexa and Co. Translators like Deepl can translate our words almost perfectly into other languages in a matter of seconds.
When surfing the Internet on a daily basis, the advertisements shown to us are selected by artificial intelligence that try to play out the most attractive product for us based on our interests and activities. We encounter these so-called “Recommendation Systems” everywhere online: Amazon, Google, Netflix, Facebook. They are a very powerful system because more and more media vie for our attention, there is more to discover online than we can ever perceive in life. Computers, therefore, have to make a preselection for us - and over time, AIs learn to understand us better and better and to play off our preferences (against us).
But also outside of the online world, AIs are finding their way into our everyday lives. Robot vacuum cleaners clean our floors and use algorithms to recognize their surroundings. Navigation systems find the optimal route. The greatest progress is currently being made by autonomous vehicles that collect millions of test kilometers on roads - even if they are still years away from widespread use.
Even if the AI has so far mainly been used by large corporations - medium-sized companies can also benefit from it. One example is wind power: the PiB research project aims to predict the icing of wind power plants .
Different types of AIs
A large number of very different technologies that have been researched over the past 70 years are collected under the umbrella term AI. The examples and procedures described so far relate to a special area of AI research, machine learning (ML). It stands for learning from experience. We have limited ourselves to this area so far, as ML is the most relevant form of AI for companies in commercial use today and much of the latest research comes from it, whether it is speech recognition (Natural Language Processing) or image processing (Deep Neural Networks). , goes.
But there are also completely different approaches. This includes so-called expert systems, which fall back on a knowledge base compiled by experts in order to draw conclusions based on certain rules - they are more or less the opposite of “learning from experience”. Probably the most famous example of an expert system is the chess computer "Deep Blue" which defeated the world chess champion Gary Kasparov in 1997.
Both approaches are often classified into different categories - symbolic and sub-symbolic AI. A symbolic AI comes to results in a comprehensible way, in which it combines symbols (i.e. words, letters, numbers, etc.) according to preprogrammed rules in order to draw a conclusion. An example of this would be the classical logic (symbol 1: "All people are mortal", symbol 2: "Socrates is a person", conclusion: "Socrates is mortal"). An expert system is a symbolic AI.
A subsymbolic AI, on the other hand, does not come to a result through the combination of symbols and rules. On the other hand, it resolves information into mathematical formulas and optimizes these formulas until they produce the desired result. In retrospect, it is not possible to follow the results directly from the formula. That is experiential learning - machine learning.
Both AI approaches are not mutually exclusive - there are efforts to unite them or to use elements of one in the other. More about this in: Understanding the difference between Symbolic AI & Non Symbolic AI.
Use artificial intelligence in the company
The use of AI for their processes is already very attractive for companies today. Companies should, therefore, ask themselves the following question: What can I achieve with AI? The first look is at your own data - which already exists in the company, which could still be recorded? AI can draw conclusions from them that were previously not possible - for example because the analysis effort would be too time-consuming for people or because there was no way to get the right answers. In this way, it can free up capacities, save manpower, or create completely new insights and enable new business models. As a cross-sectional technology, AI is relevant for every industry, as data is generated everywhere - in retail, in heavy industry, in the service sector.
What companies can expect: Once a job is found for AI, it will do it better than anyone. Because it is not only faster, due to the constantly growing wealth of experience, the error rate keeps falling. According to the company, the Google AI “Lyna” (LYmph Node Assistant) can detect breast cancer on images with a 99 percent probability , a value that doctors dream of.
It is important to find a specific application because AIs are not (yet) general problem-solving machines. One requirement would be, for example: "We want to check the quality of work parts from the assembly line in real-time using camera analysis without having to resort to manual random samples."
Like all profound innovations, the successful implementation of AI in a company also takes time. For a project to be successful, in addition to a good quality of the available data, appropriate knowledge is necessary. In addition to hiring their experts, small and medium-sized enterprises, in particular, can collaborate with cooperation partners in research projects. They carefully introduce the topic and enable you to get to know the new technology with relatively little use of resources.
Because the training of intelligent networks requires a high computing power, which can be achieved either through an investment or by renting cloud capacities - a partner who already has the capacity makes it much easier and cheaper.
So - do medium-sized companies now have to rely on AI in order to survive?
Medium-sized companies naturally find it difficult to adapt new technologies quickly. Large corporations lack the resources for experiments and the agility of start-ups without running costs.
So is it better to wait? The answer is clear: yes and no. AI technology is still young, even if it has been explored since the 1950s. The computer capacity has only been sufficient for a few years to operate AIs commercially. It is new territory and a successful medium-sized company that runs without it today will still run without it tomorrow.
For small businesses, investing in artificial intelligence is a risk. So the first question should be: How could AI increase my sales? How could AI reduce my costs and improve services? How can my customers benefit? It helps to deal with the technology to get an overview of the possibilities. Free information offers, such as those of the Mittelstands-4.0-Centers in Germany, help to accumulate knowledge. If a use case is found, an idea for a use, local partners and subsidies help to implement it.
The large cloud companies such as IBM, Google or Amazon also offer AI solutions, but these can quickly become oversized, especially since experts are still needed to implement them successfully. And skilled workers are rare, especially in the field of AI. Anyone who does not currently see a purpose for an AI should stay on the ball: One day it will come to the point that competitors rely on it and at the latest then it will be time to put a hand on the smart computer. And with the speed at which AI is currently evolving, that point will come sooner rather than later.
At the same time, the costs and the resources required for the use of AIs are falling rapidly. For several years there have been so-called frameworks that bring the basic tools with them to quickly set up your own AI networks - TensorFLow and PyTorch are the most widespread. one more piece of advice: data is the oil of AI. Those who start collecting, storing, and cataloging data today will benefit from it tomorrow.
Artificial intelligence and humans
Like many new technologies, AI also fuels fears. A famous study by the University of Oxford in 2013 analyzed that 47 percent of all US jobs are at risk from automation, a significant proportion of them from AI . Such numbers stir up fears that lead to real actions: Waymo, the Google subsidiary for automated driving, reports that their test vehicles were attacked several times with knives and stones. So is AI a threat to humans? A bitkom survey paints a mixed picture : 62 percent of Germans see AI primarily as an opportunity, 35 percent as a danger. Also a survey among managers found that 42 percent of them observed staff reservations.
The truth is somewhere in the middle. The AI will undoubtedly take over manpower from humans, and if it does, then in full - that is, no human will be needed for this one task. These are mostly tasks with a rather low fun factor, monotonous and repetitive in nature: Watching surveillance videos, answering standard queries, searching through documents.
At the same time, however, new jobs will be created, which will be supported by the innovative AI business models. People then have more time to use their manpower for new tasks because they work together with the AI. This would allow lawyers to spend more time with clients instead of searching through files for hours. It is also clear that more education is needed to prepare people for their new tasks and to give them the skills to work with AI systems.
And, to be honest, we don't really have a choice. Because the AI has long since found its way into everyday life and almost everyone is already using it today, albeit unconsciously - whether on mobile phones, for transfers or for navigation. It will be some time before we encounter AIs everywhere, but that time will come sooner rather than later, because as soon as an area benefits from AI, it will have massive advantages over its human counterparts and thus displace them from the market.
Nonetheless, it's important to talk about it and ask yourself where the ethics are in the machine. This is not just about responsibility (“Who is to blame if the machine has an accident?”), But also the question of how we want to organize work in the future.
The natural stupidity in artificial intelligence
AIs are made by humans - and are therefore subject to a natural problem: an intelligence that mimics humans is also subject to their mental limitations. One of them is bias , English for bias.
An example: In 2014, the AI experts at Amazon developed an AI that automatically evaluated and sorted application documents. To do this, they trained the neural network with applications from the past ten years. When the AI was trained, they found that the algorithm only selected those from men from among new applications. Reason: There were an above-average number of men among those previously employed, as is common in the tech industry. From this, the AIcreated the rule: Only hire men. (Source) The mistake was in the selection and preparation of the data. Ultimately, Amazon rejected the experiment, and applications were still searched manually.
The example shows that when designing artificial intelligence, people have to attach great importance to the selection of representative data - and are aware that they may already be biased by selecting and processing the data. This dilemma is not an easy one and must be considered when designing an AI. This is another reason why it is worth taking a look from the outside, working with a partner and experts in the field of AI.
After all, every AI is programmed by a human - and we know where our intelligence begins and ends.
Finally, a short summary of what artificial intelligence is:
- AI is the attempt to transfer human learning and thinking to the computer
- Strong AI, i.e. general problem-solving machines, belong to the field of science fiction, weak AI is being used more and more in today's world, whether in cell phones, in websites, social media or self-driving cars
- AIs are good wherever a lot of data can be analyzed and researched for patterns
- Machine learning is currently the most commercially important branch of AI
- AIs need data as a basis; in addition to numbers, this can also be images, videos or sounds
- AIs can process data better, more precisely and faster than humans, but they cannot understand it
- AIs are only programmed ("trained") for very specific purposes and have to be retrained for other purposes
- AIs will take over tasks from people, but at the same time also create new business areas and thus jobs
- AIs cannot understand the data; if they are fed with incorrect data, they deliver incorrect results