isleepinahammock

I'm calling the police right now to report you for the millions of crimes you could potentially commit over the next year.

I'm calling the police right now to report you for the millions of crimes you could potentially commit over the next year.

Before asking for a source for commonly known information, you should first demonstrate that you've done a modicum of searching yourself first.

Here's my compromise: we give the Flock CEO twenty to life instead of sentencing him to public flaying like he deserves.

Well in that case, you have to give the bus drivers rigorous training to reflect the cell phone policy. The bus driver absolutely should not have done that to your kid. That is not acceptable at all. However, that driver probably only did that because they knew your kid had a phone. If they instead knew that no kids in the school are allowed to have phones, they probably would never do something like that. It's a giant legal liability for the bussing company.

There's a big difference between, "I know my product will statistically increase the chance of early death across the population" and "I believe there is a very real chance my product will kill everyone."

All sorts of things increase risk. But child endangerment and similar laws are meant for very immediate and direct threats. OpenAI is building machines that they themselves believe could kill everyone. I see that as very similar to the idea of having fentanyl in the same house as young children. The fentanyl isn't a danger because it will increase the chance of the kids getting lung cancer in their sixties. The fentanyl is a danger because one dose of it could kill a kid instantly. The law addresses immediate threats, not statistical increases in mortality observable across the population.

I don't just think AI should be regulated, I think everyone at the hyperscaler companies should be criminally indicted. Every employee at OpenAI and Anthropic working on AI development should be indicted on 8 million counts of child endangerment.

The relevant statute is California 273a.

It is a crime in the state of California to allow a child to be placed in a position where great harm could come to them. For example, people have been indicted under this statute because they had drugs in the same house as young children. Even if no children are actually hurt, simply putting them in a situation where harm is likely is the crime.

It would normally be quite difficult to prove that the actions of OpenAI and others are endangering children. But the very words of these companies would be the first exhibit of the prosecution. They've openly confessed to their crimes! These very same companies and their leaders have repeatedly said that they're building machines that they themselves believe have a reasonable chance of destroying humanity. And the children of California are a subset of humanity.

Whether their claims are realistic or not, I cannot say. But I simply cannot believe that it is legal to build a machine that you yourself believe could destroy the world. We're so glazed over by the novelty of AI that we forget to treat it like any other threat. In most places it is a crime to perform some action that you believe has a realistic chance of hurting people, even if you have no intention or desire to hurt people.

Well, according to OpenAI, the machines they are building represent a very real risk to the life and safety of every child in the State of California. I see no reason that they shouldn't be charged with millions of counts of child endangerment. According to their own words, they are literally putting the lives of children at risk. That is a crime. That is not legal. And we should stop pretending it is.

There is something inherently evil about a technology that encourages you to turn over your voice to a soulless machine. Let the machine speak for you.

Only perverts buy pervert glasses.

I hate it. I recently went through a formal diagnosis process with a psychiatrist. And he used AI transcription. It was clear from the notes that it was mostly surface-level details. The notes included things I mentioned, but lacked depth and synthesis. The point of requiring doctors to write down notes is not just for record-keeping. The synthesis step is essential. And you completely lose that with an AI transcriber. Instead of summarizing with a clinical synthesis in mind, it just blindly copies down surface-level details. You end up with a long and detailed but extremely vapid session record.

We don't live perverts and their glasses.

I don’t like this “smart glasses are evil” sentiment.

That's probably because you're almost certainly not the type of person who will be victimized by pervert glasses.