Robots make music, paint, write poetry, play table tennis, play chess, help in nursing, operate, replace body parts, fly to Mars and are used in the industry. They are the result of an astonishing as well as faster development in the field of computers and artificial intelligence.
Robots are becoming more and more human in their looks and abilities, and may increasingly be referred to as our companions and assistants. In order for humans to consider them as our equals, it is necessary to implement emotions to the robots. A robot that is afraid or even feels guilty? Is that really necessary? Although it may seem absurd and unnecessary at first glance, it is the basis of any interaction. After all, communication is up to 95 percent unconscious, and if the robots are unable to understand the facial expressions and gestures of a human being, they are unable to understand emotions, feelings of guilt, or feelings of happiness, they can’t communicate with us.
The key to a future human-robot relationship is the emotional level.
If we do not succeed in developing this level, the coexistence will end as with us humans. Thus, the goal of the research should not be to program a robots as human-like as possible, but to use the additional emotional level to handle highly complex tasks. So we have to program robots with cognitive, social and emotional intelligence and give them the opportunity to evolve in a constant learning process.
Implementation of emotions in robots
How do we implement the emotions in robots? How do I teach a robot empathy? How do I manage to feel this guilt or happiness? I’ve spent a lot of time finding answers to these and other questions. I’ve spent a lot of time thinking, reading countless medicine, psychology, computer science and math books to find an answer to these questions and more. As complex as the topic may be, the answers finally provided me with the working of our brain. In order to be able to bring my idea closer to you as a reader, I have opted for a light and pictorial language – which should be easy for a father oft two kids. To guide you to the answer, I’ll take you on a little journey into your own brain.
We humans act globally on the basis of two basic systems: the punishment system and the reward system:
- The punishment system decides what people avoid avoiding danger.
- The reward system decides what they are looking for to increase their well-being. Avoiding dangers and looking for well-being is therefore the motto of the brain.
Exciting. These are all old insights, the other will say. What is innovative about it, may one think. What the hell does that have to do with robotics, will another think. The guy has a full tit, maybe you just think! Very good, – because that is the solution to my questions! The solution to all questions is the implementation of a reward and punishment system! The robot and any future artificial intelligence must feel that their positive action is being rewarded. But they also have to feel that her negative actions are being punished.
The reward system controls our actions by rewarding us with good feelings when we act in ways that make us feel good. The reward system is part of the limbic system. The limbic system consists of a network of pathways and core areas in the depths of the brain, which in addition to the amygdala (the amygdala) belong to other groups of cells in the diencephalon, which are in turn connected to parts of the cerebral cortex. The communication between the individual parts of the reward system is via the messenger dopamine, one of the so-called neurotransmitters, which transmit signals between the nerve cells. The release of the lucky messenger dopamine we perceive as a positive feeling that can bring us to action. For the further development of our theory, we can deduce that the robot is not rewarded by you with a chocolate, but with a good feeling. Ok, but how can we now implement these highly complex and also inner processing processes in a robot? For the development of our theory so far, we have used brain research and psychology – now we need computer science.
To implement a reward and punishment system for a robot, for example, we need a 7.5 gigahertz CPU with 8 cores. So we have the brain, but this is so far without input. As a next step we program the RP system (Reward and Punishment System). An RP system, that does not exist! Correct, from now yes! And what should the programming contain – just read on!
The programming contains all the information about how the reward or punishment system should work. Canceled means this, it contains the guideline when our robot rewards and when should be punished. Imagine a robot in a restaurant serving the customers. On the way to the table of the customer, this makes the tray pale or comes into contact with a customer, ultimately leads both to the case of the tray incl. the food and drinks. I know this feeling from my studies, – a bad feeling! But how we implement this sense of guilt in a robot? Now we need both the hardware and the software. Our robot has a 7.5 Gigahertz CPU with 8 cores (with only 7 cores in use) and distributing the base load of the system. Now we have to program our robot so that at the moment of dropping or jostling, 3 cores are automatically switched away. This will immediately minimize the frequency of the robot and can lead to the freezing of the system,- as in a human being at that moment! Due to our programming, the robot knows that only wrong action could lead to the shutdown of the cores or minimization of the computing power. The integrated learning process of the robot leads to the fact that he analyzes the situation and tries due to the performance drop in the future to avoid – the punishment system! Likewise, we can reward our robot. After a hard day’s work and 1000 lucky customers, there was not a single incident. By the time of the closing, our previously programmed software switches the 8 core to – what a blessing to get the power distributed across 8 cores! Pure reward …. ☺
Meanwhile, we also use robots when working with older people. As therapy robots of dementia patients or people with depressions. Especially in the field of medicine, the implementation of a rewards and punishment system would make great progress in therapy possible. Imagine a robot that uses voice and face reading to feel that the patient’s mood is tilting, performance is down, and the system is about to freeze. It is becoming clear that we can now use this technology across all sectors (education, medicine, care, safety, etc.) and be able to porgram it to the required characteristics of our business.
Of course, there are also approaches such as by the philosopher and AI expert Lola Canamero, director of the Embodied Emotion, Cognition and Interaction Lab of the University of Hertfordshire. She and her team of neuroscientists and psychologists are trying to solve the hormone-based approach. As a scientist, I find their approaches good, but must reject it critically due to the following execution. What use are the biological functions if the robot cant understand it? However, if the program was programmed with the RP system, because of his learning process, he knows that his actions decide whether or not to switch off the cores. A simple example: The robot has worked the whole day without incident, this action leads to the 8 core, which was programmed as a feeling of happiness and at the same time causes goose bumps on his implanted skin,- Dopamine Pure! 🙂 Lola Canamero sense. Because the addition of the 8 core is comparable on the technological level with a Dopamine. The robot can calculate faster, is relieved and saves a lot of energy – ultimately like us.
Finally, with the help of the hardware components and the RP system, we can help robots to feel if their situation is improving or worsening, and what feelings they can expect – they building neuronal networkIn the end, the robots know that a very specific action is associated with a certain feeling – reward or punishment. Only then technologies such as voice and face reading make sense, because the robot knows how to analyze the connected emotional level and ultimately to propose the most suitable action to the individual in this situation or to execute it as a robot himself. But before we begin to implement a reward and punishment system for all robots and to program it to our needs, we need a global code of ethics like the Digital Codex. Compliance with the Digital Codex will also be ensured by the reward and punishment system.
It becomes clear that the implementation of a reward and punishment system forms the basis of any interaction. From ethical standards to the handling of highly complex tasks.