Most people do not realize that much of what they consume as “news” is actually initiated by someone, somewhere with an agenda. (The media never tell you this because they are complicit). We have a current widely circulated instance of this, with AI being treated as a threat that could wipe out the human race, generated by dubious “whistleblowers” previously (and briefly) employed by tech firms. Some have been associated with anti-AI groups, and Chinese puppets seeking to retard America’s lead in this field. They have now been joined by executives of the leading AI companies publicly worrying about how potentially harmful their product could be. They even provide percentage projections of how likely human extinction will be, while altruistically calling for regulation of this “danger” they themselves have created.
This is highly suspect for at least two reasons. First, the call for regulation by the leading players in a field is usually predicated on using the government to seal their dominant positions off from competition from upstart firms, giving them cartel-like power. Second, they also happen to have inflated IPOs coming up that they want to maximize the value of, thus serving their own interests rather than the public interest.
All of this fear-mongering finds a credulous audience among paranoid people who have taken sci-fi too literally, while also worrying people who are more grounded. The fear centers on AI surpassing human intelligence and control, to develop itself autonomously, and then eliminate humans. Never mind that no one has bothered to explain just why AI would want to wipe out humanity when its development is designed to serve humanity. The latter is a more plausible scenario, as depicted back in 1970 by a movie called The Forbin Project, where an advanced central computer links with its Soviet counterpart to take control of war powers due to the way humans have messed everything up. The computer will then run things for the benefit of humans without all the rancor and conflict. Some people may even find this appealing.
Lost in this is the fact that even if AI exceeds human intelligence it is not the end-all that is suggested, because human beings have many other modes of perception and understanding that AI will never have. It is like saying that someone with a high IQ has it over everyone else in everything, when in fact it is limited to raw intelligence, which could be accompanied by limitations in other modes of thinking and perception or reflected in matters like social interaction.
But AI is a tool for humans, that can improve life in a myriad of ways, such as medicine and health, finding cures for cancer and other potentially fatal diseases, as well as rare diseases that affect fewer people and thus don’t get much attention. It will be able to do these things at a much faster pace, drastically reducing all the years of trial and error experimentation and testing. It will do the same in many other areas, guided by humans, and that is really the crux of it.
The real problem is what can be used for good can also be used for evil, but that is true of so many things in life. You don’t need AI to generate bioterrorism. The capability is already there, along with most other perceived threats. So the problem is really bad actors using the technology for nefarious purposes. To mitigate that, and any potential threat from hostile external forces and criminals, AI must be used to develop countermeasures and shields to maintain security, and even proactively counter any possible threat.
Importantly, there will not be some monolithic central AI, but rather a myriad of AIs operating at the individual level, emanating from a number of different sources. It will be widespread and diverse. That means that federal regulation, international agreements, monitors, oversight committees, etc. that so far have been proposed won't accomplish anything beyond impeding progress. Again the real problem is not inherently AI, but rather what people will do with it. As result they fail to address the specific segment that could actually prevent harm.
For the less likely rogue AI, or misuse of it, we already have the outlines of a solution for resolving the most feared outcomes. Isaac Asimov thought and wrote a great deal about robots, and decades ago came up with a set of Laws of Robotics that would be ingrained in the machines so as to prevent any kind of harmful action. The same idea can be applied to AI. This would involve implanting some fundamental restraints at the program or machine code level, or even within the circuits themselves. With proper consideration this would then be virtually impossible to circumvent.
Then the question is just what should be included in this essentially moral sense for AI. It is the moral sense that guides and restrains our own behavior, and something comparable could be created for AI. This could be developed by an entity drawn from people from a broad spectrum of fields such as moral guidance/ethics/religion, law, program development, math/probability experts, philosophers, circuit designers, scientists, etc. Such an undertaking ought to be able to come up with an ironclad set of inviolable principles, restraints, and guides embedded in AI that would prevent anything harmful. Furthermore, any violation would automatically invoke an unalterable kill switch. Ideally this would all be embedded in any source that is capable of producing AI.
Thus, AI is not a threat, but rather something that will uplift and enrich humanity. It will be ubiquitous and as commonplace as cell phones, and will be an integrated, routine, and normal, part of life. At some point, we may, in some sense, even merge with it on our own terms. But that’s another story.
