The Big Kahunas in Artificial Intelligence have announced that they think AI has a 10% chance of killing all humans. To prevent that, they have called on the government to create regulations for the industry. One can see the argument for regulation of almost any emerging technology. Humanity did not and does not know all the ramifications of various developments over the centuries, but some of the risks were foreseeable and often were avoided through regulation. AI, however, is different because of its potential power. Simply put, it should not be in private hands. For reasons of national and global security, nationalization of all AI is necessary. The only thing worse than government ownership in this case is private ownership. It is a matter of accountability.
Evan Hubinger, a lead alignment researcher at Anthropic, stated on social media that he believes the risk of AI killing all humans is over 10% in the coming decade. Geoffrey Hinton, often called the “Godfather of AI,” told the media that a 10% probability of human extinction is “not unreasonable”. Jacob Coxon resigned out of concern that Anthropic and OpenAI are “gambling with our lives,” and he has worked for both.
One will stipulate that the 10% figure is not pulled out of thin air (although it is). The solution is for there to be limits on what AI agents are allowed to do. They should not be allowed to do things that will result in the death of everyone on the planet. An example of this would be Isaac Asimov’s Laws of Robotics from 1942:
- A robot may not injure a human being or, through inaction, allow a human being to come to harm.
- A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
- A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
There is also a “Zeroeth Law” that developed later that says “A robot may not harm humanity, or, by inaction, allow humanity to come to harm.”
AI experts say this is not possible because there is no way to mathematically process terms like “harm,” and “humanity.” This is not true, and it merely demonstrates that people who are good with computer code are not necessarily wise as to the ways of the non-digital world (which composes almost all of creation). Here is a random title from the British Medical Journal that came up in a simple Google search “Quantifying the RR of harm to self and others from substance misuse: results from a survey of clinical experts across Scotland.” Quantification makes mathematical processing of such terms simple, and had the AI mavens passed any classes in the social sciences, they might know that. Harm can be defined as things like “death, injury and increased poverty and illness.” Humanity is simply homo sapiens. So regulation with the laws of robotics is possible.
However, there is another issue here that needs development. The AI industry says that it could destroy the human race, but they are not willing to stop because they might lose out to other companies. The reductio ad absurdum approach here is wonderful. They will not regulate themselves (which rarely works anyway) because if Company A adopts some safeguards while Company B does not, Company A might come in second in destroying the species. “If I cannot kill all humanity, I’ll be damned if I will let Elon do it.”
It is clear that the AI leaders lack the vision an education in the social sciences would have provided, and they lack the common sense to stop their self-destructive behavior because they are in competition with others equally devoid of understanding. It is time to take away their toys.
Nationalization of AI is the one way the people of the US can protect themselves. The idea of the government running AI has plenty of dystopian potential, but in a partial democracy like the US, it provides greater accountability for the industry than a private company ever could. Nor does it need to be 100% nationalized. For those who believe the private sector is best in technology (despite the evidence of going to the Moon, building nuclear weapon and developing the internet — all government projects), letting them keep a minority stake could prove useful.
Besides, if the US government is a partner, the worries of bankruptcy go away. They will be in a position to be bailed out if things go wrong. It might not be ethically correct or morally right, but as an economic incentive to play ball, it is wise.
A final point. If there is to be real regulation of the industry, it must have global rules, not national ones only. China and Russia and elsewhere must be included. And it is a damned sight easier to negotiate if a handful of companies are not whining from the sidelines. A single, nationally owned entity can be bound by treaty with a simple ratification by the US Senate.
It would have been much better to have rules and regulations set up in advance, but the Reagan Era allergy to government involvement in business remains. It is too late to do things right. Now the world must simply do what it can.