AI - Artificial intelligence simulates human intelligence with machines, especially computer systems. This includes learning (collecting data / information and rules for using the information), purposeful inference (using the rules to draw approximate or final predictions), and self-correction. Special applications of the AI are above all expert systems from all sectors, speech recognition, machine learning and customer analytics. The term artificial intelligence (AI) was coined in 1956 by John McCarthy, an American computer scientist. Today, AI is the generic term, which ranges from robot process automation (Robotic Process Automation - RPA) to the actual robotics.
AI - Artificial intelligence is the current hot topic, due among other things to big data warehouses / Big Data and the huge increase in enterprise data that companies are gathering today. The speed, size and variety of data is steadily increasing. Unlike humans, AI can more efficiently identify patterns in data, giving companies better insight into their data.
Artificial Intelligence (AI or A. I.), is a branch of computer science, which deals with the automation of intelligent behavior and machine learning. The term is not clearly distinguishable insofar as it already lacks an exact definition of "intelligence". Nevertheless, it is used in research and development.
Expert systems are specialized in very specific fields of application, as fuzzy logic is called a logic that can handle not only "yes" / "no" but also "maybe" or "yes".
It brings a whole new shift to business, market placement and overall business success. In the context of a holistic approach, AI technology supports productivity and the reduction of efforts - in addition, KI creates more scope for more demanding tasks and increases entrepreneurial growth opportunities.
These are the oldest forms of AI systems with very limited capabilities. They copy the ability of human intelligence to respond to different types of stimuli. These computers do not have memory-based capabilities. This means that such machines can not use the previously accumulated experience and do not have the ability to "learn". These machines could only be used to automatically respond to a limited set or a limited combination of inputs. They can not be used to rely on memory to improve their operations based on it. A well-known example of a reactive AI machine is IBM's Deep Blue, which defeated chess grandmaster Garry Kasparov in 1997.
Limited memory computers are machines that not only have the capabilities of purely reactive machines, but can also learn from historical data to make decisions. Nearly all applications we know fall under this category of AI. All of today's AI systems, such as those using deep learning, are trained by large amounts of training data that they store in their memory to form a reference model for solving future problems. For example, an image recognition AI is trained using thousands of images and their names to teach them to name objects that they are scanning. When an image is scanned by such an AI, it uses the training images as a reference to understand the content of the displayed image, and based on its "learning experience" new images are marked with increasing accuracy. Almost all of today's AI applications, from chatbots and virtual assistants to self-driving vehicles, are powered by AI with limited memory.
While the previous two types of AI have been in abundance, the next two types of AI are either conceptional or under development. Theory of Mind is the next level of AI systems that researchers are currently working on. An AI-based mind-level theory will be able to better understand the entities it interacts with by identifying their needs, emotions, beliefs, and thought processes. Artificial Emotional Intelligence is already an emerging industry and a focused discipline that is of great interest to leading AI researchers. However, reaching the theoretical level of AI also requires development in other areas of AI. One of the reasons is that AI machines, in order to truly understand human needs, need to perceive people as individuals whose minds can be influenced by several factors, essentially "understanding" of human beings.
This is the last stage of AI development that currently exists only hypothetically. Self-confident AI that has evolved to be so similar to the human brain that it has developed self-confidence. For any AI researcher, the ultimate goal is to create this kind of AI. Not only will this type of AI be able to understand and evoke emotions with which it interacts, but also have emotions, needs, beliefs, and possibly individual desires. For the doomsday hunters a technology with a horror scenario, the development of self-awareness can accelerate our progress as a civilization, but also lead to unforeseen disasters. Once this AI is aware of having ideas of self-preservation, it may directly or indirectly mean the end of humanity, as such unity could easily outmaneuver the intellect of any human being and make plans about humanity.
The classification system commonly used in technical language is the classification of the technology in Artificial Narrow Intelligence (ANI), Artificial General Intelligence (AGI), and Artificial Superintelligence (ASI).
This type of artificial intelligence represents all existing AI, including the most complicated and powerful AI ever created. Artificial Narrow Intelligence refers to AI systems that can perform a specific task only autonomously with human-like capabilities. These machines can not do more than what they are programmed for and therefore have a very limited or narrow range of competence. According to the aforementioned classification system, these systems correspond to all "reactive" and "limited memory". Even the most complex AI, which uses machine learning and deep learning to teach itself, falls under ANI.
Artificial General Intelligence is the ability of an AI agent to fully learn, perceive, understand, and function as a human being. These systems will be able to independently build multiple competencies and form connections and generalizations across domains, significantly reducing the time required for training. This makes AI systems as powerful as humans, simulating our multifunctional capabilities.
The development of Artificial Superintelligence is expected to be the culmination of AI research, as AGI will be by far the most powerful form of intelligence on Earth. Not only does ASI replicate the multi-faceted intelligence of humans, but it's also more powerful in all its tasks, as its storage space is overwhelmingly larger, and the data can be processed and analyzed faster to make decisions. The development of AGI and ASI will lead to a scenario most commonly referred to as singularity. And although the potential to have such powerful machines seems attractive, these machines can threaten our very existence or at least our way of life. At this time, it is difficult to imagine the state of our world as more advanced types of AI arise. However, it is clear that there is still a long way to go, as the current state of AI development is still in a rudimentary state.
As a controlling-related application example, we consider a machine learning algorithm for predicting sales figures. Before this algorithm is ready for use, it had to be trained first. For this purpose, a training data set is formed, which consists of the relevant information. Tolerant and intelligent models such as ARIMA or the prophet developed by Facebook are also unable to choose their own data sources and must be trained.
Influence variables such as weather, days of the week, vacation times and target variables, such as the sales volume must be clearly defined by us humans. Likewise the respective degree of aggregation - should be predicted on days, weeks months, preferably on customer level, customer groups or products. Data quality is of enormous importance for every analysis and only we ourselves can judge it as a human. The so-called Creed for every Data Scientist: Shit in - Shit out!
The algorithm analyzes a large number of models and selects the best one. In certain cases, the machine forecast produces astonishingly accurate results. On balance, for example, the machine-determined planned sales for the following year deviate from the human-determined sales in the following year by only 3%. A machine-determined target is not simply adopted by the planners - in corresponding submarkets, considerable, mutually compensating shifts become apparent. After a few months of project runtime, the actual trend is also close to the forecast.
The development and the consulting team of CoPlanner Software has been working on the subject of AI for several years. Today, we support customers from various industries who benefit from the application possibilities of the KI consultancy.
In the joint project, CoPlanner and Hadoco have enabled the abrasives manufacturer VSM to independently generate automated sales forecasts. Based on the transaction data from the CoPlanner software, supplemented by the machine learning algorithms of Hadoco Services. Today, the analyzes are not only faster, but also better than the assessments of the experts. That's why the combination of CoPlanner's proven experience with Hadoco's innovative artificial intelligence has now been extended internationally to VSM.