Access Denied
The bureaucracy loves control. The recent suspension of all access to Fable 5 and Mythos 5 AI models of Anthropic by the US government is a great example of that. According to Anthropic, only “verbal evidence of a potential narrow, non-universal jailbreak” has been provided by the government.
Anthropic has also voiced its disagreement with this approach of the government in the following terms:
“We are complying with the government’s legal directive and are removing access to Fable 5 and Mythos 5 for all users. However, we disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”
Bureaucratic Control
A few years ago, when AI was at a nascent stage, there were hardly any regulations or bureaucratic clutches to control it. But today, governments all over the world have realized that AI is so powerful that it can disrupt their hegemonies and disturb the power balance that they have been enjoying unbridled so dearly.
In one of my earlier articles, I discussed that in the famous Foundation series by Isaac Asimov, the galactic empire, spanning millions of planets, is depicted as a stagnant and over-centralized system that is crippled by its bureaucracy and excessive legislation. Today, we are entering the same era. AI will be able to do every conceivable thing in the future. This will irritate a lot of bureaucrats because they love to move things slowly. Their excessively legislated minds cannot follow the iterative speed that AI or its researchers are capable of. So, what is the best way to control AI? Control its development and do not let it develop beyond a point. And even if it does, just shut off access of the most advanced models to the general public and use them for their own bureaucratic and “administrative” purposes.
The world has suffered far too much in the name of “administration.” Excessive legislation and overcomplicated policies are like dinosaurs that will eat our civilization alive. The bureaucrats and the politicians may think that they can control the growth of AI, but in reality, they cannot. As Ray Kurzweil puts it, technology follows its own exponential growth curve that is quite distinct from the growth curves observed in biological species.
Today, it’s Anthropic, whose models have been shut down by a government. Tomorrow, there will be other companies that will be affected by such policies. I am not saying that all regulation is bad. We do need regulations to tell us where exactly we are wrong, but moral policing in the name of regulation should be detested.
Don’t Panic!
We, the common people, think that governments exist for our collective benefit. But is it really the case? Governments seem to exist for their own good, and most of the time, they are able to create brilliant facades that look like collective welfare but are not.
In the end, I want to say that governments do need to legislate and regulate the AI sphere, but the approach should be objective, the standards adopted should be scientific and transparent. Otherwise, bad actors might continue to develop dangerous or fatal AI in a covert manner, and it might be too late to catch them.