Recently, songwriters in the USA called for protection against plagiarism by AI. Others argue that service providers are liable for the statements made by chatbots if they publish their AI results. If an AI generates an image, it is not subject to copyright in the USA. In Germany, on the other hand, AI-generated images are also subject to copyright law. What all this has in common is that AI is seen as something controllable, somehow extraterrestrial. However, there is one main reason why AI cannot be regulated.
Introduction
It all started when a product called ChatGPT caused a stir in March 2023. After a minute, everyone understood that this chatbot was intelligent and that nothing like it had ever existed before.
Unfortunately, many people do not want to take note of the statement in the previous sentence. They would mistakenly say: "ChatGPT is artificially intelligent". The correct statement is: "ChatGPT is intelligent". The point here is not to convince everyone that this is the case. In just a few years, reality will replace belief or disbelief.
Currently companies and investors like OpenAI, Microsoft, Elon Musk, Nvidia or Amazon invest thousands of billions of dollars in AI technologies! We're talking about billions of dollars, not millions, which would already be a lot. It's actually about gigantic chip factories and robot manufacturers.
Why is that so? Because it's a no-brainer to put an intelligent computer on something movable that roams around in the area. This movable thing is also called Robot. Figure AI is a company that has such products in its portfolio and is investing heavily into them right now. I myself had already started investing in GPU- and Rbobotikunternehmen before anyone's article or opinion reached me.
It's no coincidence that Jonas Andrulis, CEO of Aleph Alpha, quoted the phrase "GPU poor" in a TV documentary, which he claims is popular in Silicon Valley. GPU stands for Graphics Processing Units, which are currently essential for AI calculations. These GPUs can be found primarily in graphics cards from one manufacturer, namely Nvidia. Other manufacturers are effectively irrelevant, as shown by a glance at the most popular AI frameworks, all of which support CUDA. CUDA is Nvidia's standard to command calculations on the graphics card. That AI graphics cards are no longer used for outputting images on monitors should only be mentioned in passing.
Back to the core of this article: Why can't AI be regulated?
The answer is given in the next section.
Artificial Intelligence is (artificial) intelligence
Due to inadequate definitions of Artificial Intelligence, such as those found in the EU's AI Regulation and at the OECD, a new definition of AI has emerged. This definition reads:
As Artificial Intelligence a artificial system is called that tries, a problem also with vague specification on a not precisely specified, solution-oriented way to solve and combines existing with new knowledge and draws conclusions.
Source: Klaus Meffert in Dr. GDPR Blog
The beauty of this definition is that the definition of intelligence can easily be derived from it. In contrast, the EU and OECD definitions fail to do so.
The definition of intelligence is:
As Intelligence, a system is referred to that attempts to solve a Problem even with vague specification in a non-concretely specified, solution-oriented way to solve it and combines existing knowledge with new one and draws conclusions.
Source: Klaus Meffert ib Dr. GDPR Blog
The definitions of artificial intelligence and intelligence are virtually identical, which is correct. They differ in exactly one word. This word is "artificial".
It doesn't matter what form an intelligent system takes. That intelligence has just as much or little to do with statistics as everything else has already been discussed. Brief summary: Everything is Statistics, see Quantum Physics or Heisenberg's Uncertainty Principle or the radioactive decay of particles. Statistics is not a unique feature of AI and can be found in the brain of humans but also elsewhere.
Digression
Researchers have apparently succeeded in deciphering brain patterns of humans (or animals?). The approach is outlined as follows:
The process uses AI to establish a correlation between measured brain impulses and inputs and outputs that the test subject has received or made. If that's not intelligent…!
The procedures depicted in the image are all common. Revolutionary is the approach of combining them in a clever way.
Can intelligence be regulated?
Intelligence can be regulated, namely through prohibitions. However, these bans must be effective if they are to actually regulate something. A ban is effective if
1) a violation of the prohibition can be established and if at the same time
2) a violation of the ban is consistently punished.
Whether laws that prohibit certain actions are consistently enforced, largely depends on societal acceptance and also the opinion of judges, whether a law is truly sensible. If a judge thinks that then everyone could come if rights are broken, then the court would be overwhelmed by plaintiffs. So many a judge will not give the plaintiffs right, although they (from a theoretical objective point of view) are in the right. This statement is based on evidence.
My name is Klaus Meffert. I have a doctorate in computer science and have been working professionally and practically with information technology for over 30 years. I also work as an expert in IT & data protection. I achieve my results by looking at technology and law. This seems absolutely essential to me when it comes to digital data protection. My company, IT Logic GmbH, also offers consulting and development of optimized and secure AI solutions.
