Neuromorphic computer technology - beyond the current AI
How chips emulate the interactions of the brain
Neuromorphic computing goes beyond the limits of the previously weak AI. The new chips emulate the neuronal structures of the human brain and dramatically accelerate learning processes.
As early as 2017, a major semiconductor company introduced the first research chip for neuromorphic computing, enabling scientists to experiment with dramatically accelerated learning processes, which the chip made possible. This chip had 130,000 neurons, which corresponded roughly to the abilities of a ladybird. Eight years later in 2025, a further development of this computer processor with two billion neurons goes into series production and will change the world in medicine, autonomous driving, air traffic control and many other areas of life.
But what distinguishes the ability of the novel chips with artificial nerve cells such as neurons and synapses from the usual chip technology? The information technology that we knew about until 2020 and that was built into smartphones and PCs was based on the principle that John von Neumann designed in the 1940s. This architecture of control, calculation and memory proved to be effective over many decades. Computers could be programmed with anything as long as these were separate logical processes based on zero and one.
However, this classical architecture is not suitable for extremely fast learning processes and the ability to understand, process and interpret uncertainties with a resulting independent decision. But this is exactly what is needed to achieve processes and decisions with artificial intelligence. Artificial neural systems offer the possibility that thousands of neurons can communicate with thousands of other neurons simultaneously, process and store information.
From 2025, the first neuromorphic computer systems for medicine are being created. In the future, radiologists will be supported by these systems in image evaluation. They already analyse tissue structures during imaging and prioritize which images and tissue areas the radiologist should focus on by means of uncertainty measurements. Another example is the AI required for autonomous driving with vehicles. Conventional computing is suitable for many processes, such as navigation along a GPS route. However, neuromorphic systems are far superior in the evaluation of sensor data such as GPS or cameras that, for example, detect a ball rolling onto the road but which is then covered by vehicles. Only they will be able to deal with the uncertainties in the interpretation of an uncertain situation and act like an experienced driver. Around 2025, many large technology and industrial groups are increasingly relying on neuromorphic computer technology and create a new hype in which special copper alloys also play a role - whether for voice recognition, autonomous machine inspections or for optimizing data and Internet security.