You lot know you can turn it off, right?

If you have read the news recently you’ve probably seen a load of articles telling you that we’re all gonna be dead before the decade’s out. The reason for humanity’s latest extinction event is… AI. In the latest episode of ‘AI: how to make a moat’ we’re seeing members of Anthropic leave their job to sound the alarm bells to the world.

According to them, execs at Anthropic have been swimming in digital hallucinogens for so long now that they are actively terrified of the potential that their token incineration unit beholds. It seems as though they are completely unaware of the fact that it’s a fucking machine which requires an active internet connection and a source of electricity.

They are not alone, since the inception of these energy guzzling word faucets our extinction has been thrown around as a possible side effect of their advancement. A small price to pay if you ask me. Nonetheless, this extreme fear being rammed down our throats does beg a simple question: have people forgotten how computers and existing work?

The simple fact of the matter is that LLMs need computers, computers need electricity, electricity is generated by machines installed by humans… you see where I’m going with this. Humans on the other hand need very few things, none of which are electricity or computers. In fact, you could argue that everything which a human truly needs to exist can be found for free exclusively in places where there are no computers.

What I’m basically trying to say is that the dependency tree for a human being’s survival does not include compute (it does include trees, ironic) while the dependency tree for AI does include a human not turning off the power.

With that in mind it is fair to presume the only way AI would end up killing us all is if there were some nut-job with a fragile ego who is also dumb enough to make an autonomous nuke in charge of a nuclear power… oh, wait.

In all seriousness, though. It is pretty clear what is happening here. The ‘frontier’ labs are beginning to feel cheaper, open, models nipping at their heels and realising they have no moat. How could they possibly get their moat? Regulation. If the frontier labs are able to convince western governments that “AI dangerous, China AI bad” they will be able to utilise the red tape of regulation to make it incredibly difficult (and possibly illegal) to use anything which has not been signed off by their buddies in the government.

Obviously, I could just be looking at this all wrong. A recent article from the BBC did include quotes from Sam Altman where he stresses that the world should trust our tech overlords and that they are “going to do the right thing because it’s the right thing”. I mean, why would he lie? It’s not as if he has been labelled as a person who exhibits a “consistent pattern of lying”.