Please regulate my industry
The title of this blog, "Please regulate my industry", has been said by roughly no one, ever, except when the speaker in question thought that they could solidify their position in the market, and harden their position, by "accepting" regulation that in most cases they helped to promulgate. When you read about OpenAI or Anthropic or other AI companies arguing that the government should step in and regulate their industry, you should think: 1) well, AI could be pretty dangerous 2) why would government bureaucrats be better at containing the danger than the creators of AI 3) why would any private enterprise actively seek regulation unless the regulation gave them an advantage over other potential competitors?
You can both think AI is a potential danger to society AND question the motivations of the executives who clamor for regulation. And that's exactly what I think we should do - question the motivation and ask quo bene? Who benefits?
We all do
If you believe that AI is literally out of control, I have news for you: the horse has already left the barn. As with all new technologies, science and technology progress far faster than we have the capability to apply philosophy, reasoning and political or legal safeguards. People who are crying wolf now are probably six months too late. So, trying to close the barn door may slow other dangerous or deadly AIs, if they are out to get us, but there's a good chance that if the executives are right, we are already at risk.
That does not mean we should not be concerned, it just means we should ask - what regulations or oversight? How does a government or bureaucracy determine when an AI becomes sentient, and more importantly, sentient and angry or uncontrolled enough to end life on the planet? Did none of the executives read Asimov and implement the three rules for their code?
All government executives can do now is to listen to the worries and attempt to protect the population from implied potential threats. After months of hearings, years of regulation writing and several court cases, we will all live in the land of the Terminator if there is significant risk before the bureaucracy gets put into place.
If, on the other hand, the executives are telling us that they have released the genie from the bottle, then they should be invited to step down, disconnect all the AIs that are dangerous and allow us to go back to the abacus or other tactile adding devices. If they created the AIs and have lost control, shouldn't they be the ones to make critical decisions about how to manage them? Would you rather have top AI scientists acting immediately, or the first 50 people in the Boston phone book developing a committee to study this? And, if the executives simply cannot or will not curtail dangerous technology, then we should put them in prison. That's what we do with other executives who create faulty products that damage the lives or livelihood of others, and most of those executives don't claim to be responsible.
On the other hand
What if the executives of Open AI, Anthropic and others are raising warning flags and asking for regulation to protect themselves and their companies from new entrants or foreign competition? Few industries ever invite regulation. As Reagan once pointed out, some of the most feared words are: I'm here from the government to help". Other than in healthcare, the military or FEMA, I'm fairly certain that those words strike fear into most businesses.
There is a rationale for asking for regulation, and that's when you've grabbed all the market share or position or investor value that you think you can, and you are trying to pin all of that value in place and develop defenses against other competitors or people who could sue. Once you've helped define the regulations (because who else has enough experience or is keeping the wolf at bay?) then the regulations become onerous for others, but a defensive moat for yourself.
Free market economy
A free market in an ideal world would have no regulations but would have tremendous side effects and externalities. For instance, people would dump chemicals into rivers until they catch on fire. We create regulations around specific activities or industries specifically because companies in an unregulated free market would dump pollution just about anywhere and might be tempted to create products that were untested or unsafe. While people who lean libertarian will talk all day about free markets, most will admit that some regulation is a good thing, or they've never heard of Love Canal.
But regulation needs insight to the risks or the industry and a balance between the profit motive and the desire to keep people and the environment safe. The current situation with AI and regulation is attempting to balance Armageddon on one hand combined with outrageous profits and valuations with significant regulations on the other hand, with little capability or experience on the part of the regulator.
Physician, heal thyself
Perhaps we should first ask the AI companies themselves to ensure their models aren't threats to people, physically, socially, psychologically or otherwise. That expectation is no different than what we expect from makers of automobiles, blenders or shotguns. Used carefully and respectfully, all three of those objects are safe and useful. Used dangerously, all three could main or kill uses or other individuals.
If the AI executives cannot guarantee the safety and security of their users or the population as a whole, then perhaps we should take the more dramatic action and shut them down. I'd like to see what would happen if this proposal was made in earnest.
Comments
Post a Comment