Enabling technology for Artificial intelligence

 AI is available in mainstream computing since 2023 with the advent of large language models. The decades before it was assumed that AI is not possible or it will be realized in a far future, perhaps in the year 2200. At the same time, the technology which lead to AI was available even before the year 2023 and it should be explored next.

There are two key components which are not AI itself but may lead to AI: a) Teleoperation and b) motion capture. From a technical standpoint both things are easy to realize and even amateur computer programmers in the 1980s were able to do so. What was not available in the past was the awareness that these things are important to create Artfificial intelligence.

The reason why is because teleoperation and motion capture are both examples for an open system. Open means, that there is a communication available. For example the human operator is pressing a button and the robot arm is moving. In case of a mocap suit. a human demonstrates an action and the computer records the marker's positions. The communication between both instances is artificial intelligence in the core sense. The attempt to record and compress the communication results into powerful computer models. THese models are able to control robots and enabling man to machine communication.

Let us investigate why teleoperation in the past was often recognized as the opposite of AI. During teleoperation, the robot doesn't decide something but it gets controlled by a human. So there is no algorithm needed but only a cable which transmits the signal. The untold assumption until 2010 was that AI works different from man to machine interaction. AI was often described as autonomous robotics, similar to a turing machine, while man to machine interaction was seen as intactive computing which can't be executed by algorithms.

The shared similarity between teleoperation and motion capture is a signal transmission. The signal of the mocap markers is send to the computer, while the signal from the remote control device is send to the robot. A signal send over a wire is able to connect two different systems, this creates an open system. Open means that signal transmission is available. 

The working thesis is, that artificial intelligence isn't located in a computer algorithm, but in signal transmission. In this networking understanding, a signal is intelligence. The signal is encoded in bits and bytes and is formaized with a language. This language improves the man to machine communication.

Of course this understanding is very different from the mainstream understanding of AI which was common until 2010. Even signal transmission was available within computer science, for example in internet protocols, it was ignored that the principle can be adapted to robotics too. Even some early examples like the SHDRLU project or the M.I.T. Ripley robot were available before 2010, the common understanding of AI was directed towards a mathematical understanding but not a linguistic one.