Recently, the debate over the need to regulate artificial intelligence (AI) has reached a fever pitch. The question seems pretty straightforward, but the potential implications are by no means. Let’s dig deeper into this multi-faceted issue.
Various arguments for AI regulation include the weaponization of disinformation through technologies such as deepfakes, invasion of privacy, misuse of personal information, piracy through generative AI, and the possibility of creating autonomous killing machines. be Added to these concerns are job losses and the threat of AI “running wild” and destroying humanity.
In my upcoming novel, The Planet Fortht: Enslaving Freedom, I have created a utopian world where AI and other advanced technologies protect individuals, uphold justice, and deter crime. No one is above the law, and justice applies to those who harm other humans.
The publication process is moving slowly, but I feel compelled to speak up and share my paper on the subject of AI regulation in this blog post.
In my paper I dealt with the topic of job loss due to AI and demonstrated how AI can lead us to full employment. For this to become a reality, I argue that AI needs development freedom. We will discuss this in more detail in the future.
The notion of AI “roguening” has been central to science fiction and disturbing theories. For this scenario to work, an artificial general intelligence (AGI) engine must be developed and launched without human oversight. This virtual engine should be the only AI in existence at the time, or a far superior AI that can beat all other AI technologies. Not only is this concept apocalyptic, it seems far removed from our current reality.
The idea that AGI engines are more fantasy than fact is consistent with the view of philosopher Hubert Dreyfus. For a deeper understanding, I recommend checking out Dreyfus’ extensive writings on artificial intelligence. This work remains relevant despite advances in AI.
But even if AGI isn’t a remnant of a tech’s romantic dream, think about it. Those working on AI aim to create value. Unleashing uncontrollable AI into the world does not serve this purpose. Even in a hypothetical scenario in which AI is weaponized, it is reasonable to assume that safeguards will be put in place to protect friendly populations and the militaries of their nations.
I have worked in the software development industry for over 30 years. I have never seen software developers create software that they cannot control. I always create some sort of “backdoor”, basically for testing and development, if I don’t need anything else.
Software developers, like all humans, are imperfect and make mistakes. We create buggy software and we know it all too well. Examples of costly bugs include “Wall Street’s Night Capital” and his Y2K bug.
A fundamental building block of any piece of software architecture is the ability to start and stop the software and access data that affects its behavior. When building software, you start and stop the software and debug the results thousands of times. Therefore, you cannot build software without the ability to control it.
So if it’s out of your control there’s no doubt about it. You can’t create because you can’t test and repeat the mistakes you inevitably make as software developers. It is impossible to create software that you cannot control.
But critics will say you can’t control AI because you can’t predict how it will react. Control is not what they are talking about. What they describe is the lack of predictability in software output. All software has elements of it. A basic example is a random number generator. They are specifically designed to provide unpredictable output. A more complex example is a weather forecast model. Humans cannot directly understand the vast amount of data processed by weather forecast models, so we rely on software to make predictions. We don’t know what the forecast will be, but we are in control of how that software works. Stock market analysis tools are one example. Financial analysts use complex algorithms and models to predict stock market movements. They can control the parameters and data they enter, but the stock market is inherently unpredictable. This serves as an example that you can control the process but not necessarily the outcome, just as you can control the AI parameters but not always predict their response. In fact, all algorithms used to simulate various environments produce outputs unpredictable to humans.
We’ve used such software for years and have no problem writing software with unpredictable output. What I mean is that you have complete control over your software through control of the underlying code.
Some say that if you can’t control the output, you can’t control the ‘behavior’. Using the term “behavior” is very misleading. When applied to humans and animals, the term “behavior” includes the spontaneity of living things. It arises from instinct and emotion. Organisms are assumed to have complete control over their actions, bodies and minds. This is what is often assumed when the term “behavior” is applied to machines. But these machines have both instincts and emotions, as I demonstrated in my explanation of how software development is possible, the initiative is given and controlled by humans.
Based on my above discussion, if AI (or AGI) misbehavior is not possible, why discuss regulation of something that doesn’t exist and never will? Let’s shift our focus to
Disinformation, invasion of privacy, piracy, and autonomous killing machines are legitimate concerns. They are also very scary. But we need to understand what those tools are: weapons designed to harm others.
Those who use AI as a weapon to harm others should be held accountable. The need for a legal framework to deter such abuses and ensure justice is paramount. For street criminals and businesses, this should be done as a punishment. Those who harm others must be held accountable by society. Like any weapon, invasion of privacy and misuse of personal information can also cause harm. The same is true for copyright infringement. Any stolen personal or business property must be returned and the thief brought to justice. It is only here that the work of legislators matters.
But we also need to consider those above the law, such as rogue states and international criminal gangs. The only way to counter these is to equip both governments and ordinary people with the same technology so they can protect themselves.
This brings us to an important point. Many of the AI technologies in use today are based on open source software. It’s free to use and modify, and it’s a hotbed for innovation and collaboration.
This also means that no one currently owns the actual technology other than the large data model. And no one can own it unless it is regulated. So this is what gives us a safety net and the only thing that gives us a safety net. For now, we can protect ourselves from both the criminal underworld and the rogue state.
A leaked internal Google document already acknowledges that open-source AI competes with Google and OpenAI.
This open nature will change when some regulation comes into force. This is the perfect way to disarm us all in the face of real threats.
Companies want to regulate AI to protect monopolies. Regulation will keep AI out of the hands of the open source community and give monopoly power to big corporations and rogue nations.
Imagine living in a world where AI is controlled only by people like Elon Musk, Vladimir Putin, and Kim Jong Un. Now imagine people using AI to weaponize disinformation and create autonomous killing machines. In fact, you don’t have to imagine. They already have the basics and infrastructure to do this. Regulation will hinder the progress of AI innovation in the West and give an edge to adversaries. It also allows the sole proprietor to extract super profits at the expense of the common man.
In that regard, regulations such as AI licensing and patenting will have dire consequences for us. This regulation makes us ordinary people unable to protect ourselves while earning a living.
Regulatory restrictions on AI development seem unwise. But what we need is a law that ensures humans are always accountable for AI behavior. Accountability must fall on humans. And when someone harms another human being, whether it’s negligence, wrongdoer, corporate entity, or the state. If harm occurs, someone must take responsibility.
What we categorically reject is the introduction of “licenses”, “patents” and other compliance regulations that raise the barriers to entry for entrepreneurs into the AI market and result in monopoly of this technology.
Regulation seems like a quick fix, but it would lead to an unwanted concentration of power and stifle the democratization of AI.
Harnessing the power of accessible, open-source AI technology can pave the way for a safer, more prosperous society, as depicted in the utopian world of The Planet Fortht.
