DMT is naturally produced in your brain. Even though you might think a lot of people are "insane", DMT does not cause "insanity".

Ayahuasca is DMT, it's a drink that combines DMT containing elements with other elements that allow the DMT to be metabolized in the stomach instead of being destroyed upon ingestion. DMT can also be inhaled by itself, or injected as in the case of this story. DMT is also naturally produced inside your brain.

DMT is also naturally produced in your brain.

> Altman called the commercials "deceptive" and "clearly dishonest."

So pretty much like the output from his own ChatGPT.

> And Google's version at the time didn't have anywhere NEAR this privacy-raping capability.

I happen to know one of the main developers of Google's facial recognition systems, and way back in the day probably 20 years ago, before "Google Glass" was a thing, someone had the idea of being able to identify random people out in public using Google's tech. The developers shut down that idea quickly. That was just too much for them. Facebook is vastly different than Google was back in the day, so it comes as no surprise that Facebook would be the ones pushing this tech. I'm just surprised they didn't get to it sooner.

All this work they're doing so we can get lied to constantly by a machine.

Apple has a small market share for laptops and desktops, making up only 15% of the market, and has stayed there for decades. Most people can't afford 5k or 6k displays, and won't spend for an 8k display. There's no real market for these things outside of a few wealthy fanbois.

It seems like you don't realize that not everyone that did business with Epstein was a pedophile. Epstein had half a billion dollars when he died, and he did use that money for legitimate business reasons, with people that did not participate in his pedophilic tendencies. Trying to cancel everyone and anyone that ever had any contact with Epstein just makes you look crazy.

Slop is slop, whether machine or human generated. And unfortunately, I can always count on the machine to generate slop. That can't be said for all humans, because at least I can fire a human that gives me slop, but the machine is what it is - if you trust the machine more than a human, then you are a fool.

What happens when people can buy premade 3D-printed gun parts directly from China? Will they ban all 3D printed parts from outside the US? I'm not sure the politicians have really thought this through. Anyone can order any kind of part to be 3D printed from China and have it delivered within a few days.

They will just ban all 3D printers from China, the same way they just banned all DJI drones (which are made in China).

I'm an "AI" skeptic, but yesterday I kind of had a win with Claude.ai. I needed to create a server that implemented a couple dozen APIs for a 3rd party service we're evaluating. The 3rd party service connects to these APIs to ingest data into its system. Since we're just evaluating this 3rd party service, I didn't really want to invest too much time writing APIs for it. I gave Claude.ai some clear instructions, and pointed it to docs for every API I needed. I also specified that wanted the API server set up as an AWS Lambda using the latest nodejs version.

The first attempt did not work, but I noticed that the "index.js" file was incorrect. It did not contain a "handler" function that the Lambda system requires, but that was in a separate file "routes.js" which was not referenced anywhere in the "index.js" file.

Deleting the index.js file and renaming routes.js file to index.js made it work. So, Claude.ai made a stupid mistake, but all the rest of the system worked. The couple dozen API files were correct, the example data was correct, and it actually worked once I pointed out the mistake to Claude.ai and it created a new .zip file of the project.

But I wasted a lot of my time trying to figure out why it wasn't working, and setting up API Gateway was another problem I had to figure out. It took me a few hours to get what Claude.ai wrote to work, which is about as much time as it would have taken me to just do it all myself.

Claude.ai is not really getting much better, but I'm getting better at expecting how Claude.ai is going to mess things up.

The downside of using Claude.ai is that I don't really know what it wrote unless I look through a few dozen files. I don't "own" that code, I don't feel invested in that code, but I could probably expand on it and eventually get familiar with it.

Generative AI is like using tracing paper. It will use existing images wholly or in part to generate a result that satisfies the query. A human artist asked to draw a bridge can create an image of a bridge that never existed before, purely out of imagination based on walking the earth and seeing bridges. They aren't copying anything, they are using their own experience, and they can draw the bridge from any perspective in any way they want to. An "AI" can't do that. The only thing an "AI" can do is regurgitate the "training" data. It takes a photo of a bridge that may or may not be copyrighted and uses that to create the output. It does not have an imagination, it does not have talent, all it has is original content it ingested from copyrighted and non-copyrighted sources that it did not itself experience.

> "The entire American economy right now is one big bet that that's going to happen,"

So what was "the entire American economy" before "AI" started to happen? Is that just all gone now? So the American economy is based now on nothing more than "AI"? This entire line of thinking is ridiculous. No the "entire American economy" is not based on "AI". There's far more to an economy than the latest tech buzzword.

Kind of funny you say that - I've always called MacOS and System 1 through 9 before it a "Fisher Price Operating System". So much about it is just so bad.