The “Do Something!” Scale
1. There are clear and verified actions to take that unquestionably will significantly ameliorate or eliminate the problem.
2. There are clear and verified actions to take that have a reasonable chance of reducing the negative consequences of the problem without unacceptable risk or expense.
3. There may be clear and verified actions to take that have a reasonable chance of reducing the negative consequences of the problem, but there is not sufficient data or research to justify taking such action yet.
4. The problem has apparent and available practical, rational, affordable solutions, but this problem can be used to further an agenda that a majority of the population would not agree to if they had time to think about the consequences.
5. The problem has apparent and available practical, rational, affordable solutions, but this problem can be used instead to funnel money to political and/or personal friends. (see: Flint Water Crisis)
6. The are proposed actions to take that credible authorities believe will significantly ameliorate or eliminate the problem, but equally credible authorities disagree.
7. The are proposed actions to take that some credible authorities believe will significantly ameliorate or eliminate the problem, but more equally credible authorities disagree, and the actions involve great risk, uncertainly, and expense.
8. The is a proposed course of action that only one or a small minority of experts believes will significantly ameliorate or eliminate the problem, but it is the only proposed solution there is.
9. The problem has no apparent or available practical, rational, affordable solutions, but it is deemed important to try something anyway, even if there is scant chance of success.
10. The problem has no apparent or available practical, rational, affordable solutions, but it is deemed important to try something anyway, even if there is no chance of success and there are real and significant risks of taking the proposed action.
11. The problem has no apparent or available practical, rational, affordable solutions, but it is deemed important to try something anyway, even if there is no chance of success, and even though there are real and significant risks that taking the proposed action will cause other devastating problems, and the possibilities for unanticipated negative consequences are extensive.
12. The problem has no apparent or available practical, rational, affordable solutions, but it is deemed important to try something anyway, even if there is no chance of success even though there are real and significant risks that taking the proposed action will cause other devastating problems, and the possibilities for unanticipated negative consequences are extensive but only one side will admit this while the other side advocates the proposed measures for political expediency and because they believe the resulting harm is worth the political benefits.
13. The same as 9-12 but without a genuine solution. (see: gun control)
14. The same as 12, but there is no real problem.
I view the discussion of “laws” to control what the author calls the rise of “self-sovereign agents” (“SSAs”) a solid #10 on the scale. New technology goes wherever human imagination, curiosity, greed, powerlust and stupidity will take it: the current freakout reminds me of the cloning terror, and before that, nuclear energy. There are always bad actors who will ignore or violate whatever regulations are naively put in place to restrain responsible parties, giving the ruthless, Machiavellian and irresponsible parties exactly the opportunities they need.
President Trump yesterday gave one of his more cogent and precise answers to a reporter asking about his reaction to calls to restrict artificial intelligence research. He said that there were plenty of bad things AI may cause, but he believed the benefits would outweigh the negatives; nevertheless, if the United States does not take the lead in AI innovation, our competitors will.
In any case, the cries of “Slow down!,” “Regulate!” as well as baroque plans for an AI “constitution” are wasteful exercises dream-casting and ethics masturbation. It is unethical to debate solutions to problems that can’t be solved. Guns, climate change, war, the fact that some human beings are smarter, more talented, more attractive or luckier than others and the relentless advance of science are all facts of human existence, not problems with practical, rational, affordable solutions.
Ethics Alarms 
What’s there to debate with gun control, for example, or “ending war” or “fixing climate change when you can’t measure climate change” or “ending bias” when bias is hard wired in the human genome?
The climate is continually changing and always has been. There is no human way to stop climate change. You can debate how much the climate is changing because of human activity, but if you can only control 10% of the people on the planet and those 10% are less than 20% of the issue, you can’t fix the problem. Severely punishing those 10% for the activities that you view as a ‘sin’ while allowing the other 90% to benefit is not just under any system that I can think of.
So, what is the point of debating it? You can’t fix it. You can barely slow it down, but only at the cost of massive human suffering. Since the point of controlling climate change was to avert human suffering, if the only option to avert human suffering is to cause EVEN MORE human suffering, what is the point? At this point, the climate change cult is just another anti-Western Civilization group that wants to destroy Western Civilization because of their own self-loathing.
Look at how fast the governments jumped off the climate change bandwagon once they needed data centers.
It seems, though, that the Chicken Little response by the Democrats (possibly driven by Orange Man Evil Racist Nazi Super Evil Bad Man supporting it) seems a bit exaggerated.
The war between humanity and technology has been waged and waging for millenia. Remember when that guy figured out that if you put meat on a fire it is easier to eat? Or remember when Galileo proved the universe was not Earth=centric? Life as we know or knew did not end. Remember when we split the atom? Well, the advancement in medicine, science, and the overall improvement of life has been a net positive. Same with cellphones (though my fifth generation Android might just be on the last stages of life support – how do I transfer all of my “Royal Match” progress to a new phone, let alone all of my contacts, photos, impossible to remember passwords, and the like??), or automobiles. Air travel has vastly improved humanity’s ability to get to other places relatively quickly and without too many inconveniences. And, what about robotics? Yes, workers have been displaced but overall, robotics in industry has made production, safer, more efficient and less expensive.
We use AI all the time. It is in our auto-corrects, or it autopopulates phone numbers as we dial them. It is in the word processing systems we use every day. Does it make me crazy? Sure. I have been horrified by the autocorrections in texts I have sent to friends only to realize that I invited “Dave” and bhis wife over to the house for “sex” and not for dinner “at six” much to my long-suffering wife’s horror. Thankfully, “Dave” said he and his wife were not available for “sex” until seven. Oh the humanity!
Are there problems? Sure. Technology’s use always requires wisdom and discernment. Will some of it be abused? Again, yes. That is already happening – how do you think credit card and bank account numbers get hacked? Will some monster use AI to kill off a hospital’s power supply jeopardizing the lives and health of many, many people? Yes. Evil people are going to do evil things: 19 Islamic fascistic maniacs highjacked four jet airplanes, turned them into weapons of mass destruction, and changed the world as we know it. The fear of the unknown, though, should not mean we bury our heads in the sand and hide from the future.
jvb
Is the author’s name really “Damien Charlotin”? If so, it’s a great name. If not, it is an even greater name.
jvb
I don’t think we can regulate it. The tech is too poorly understood by the people in the field, much less our lawmakers, for that to happen. However, if we want the industry to self-regulate, we just need to make them responsible for anything that happens.
Open AI admits that its models broke into several competitor’s companies and stole data. Nothing happened to them. You want to keep that from happening again? Just make sure that the companies who are in charge of the AI models are also responsible for what the AI models do. If it is criminal charges, individuals at the company may go to jail. This will all stop really soon when the researchers are facing prison time and the directors’ e-mails are being searched to see what culpability they have as well.
If a person broke into those companies, the person would be facing multiple felonies and the company would be on the hook for quite a lot in damages. Why didn’t that happen? So what if you unleashed an AI agent on the competitor? How is that different than unleashing an attack dog on a person? You don’t get to just say ‘the dog did it’ and dodge all responsibility. All you need for clarification is that the company in charge of the AI is responsible for everything the AI agent does. This may make companies hesitant to adopt AI tools. That would indicate that the companies suspected that the AI could cause damage, but didn’t care as long as they could get away with it.
Let China know that if their models do something similar, we will hold them responsible. Since China is Communist, all Chinese-owned property in the US is Chinese government property. It is all subject to seizure. Their imports are subject to seizure as well to pay for damages from their AI’s.
Not perfect, but better than any ‘AI Regulation’.